ist project # 2001-32603 - 6net · brief description of the product (brief functionality etc)...

IST-2000-32603 Deliverable D

Project Number: IST-2001-32603

Project Title: 6NET

CEC Deliverable Number: 32603/PSNC/DS/6.2.3/A1

Contractual Date of Delivery to the CEC: June 2003

Actual Date of Delivery to the CEC: July 23, 2003

Title of Deliverable: Interim report on development and test

Work package contributing to Deliverable: 6

Type of Deliverable*: R

Deliverable Security Class**: PU

Editors: Bartosz Gajda, Wiktor Procyk

Contributors: Isabelle Astic, Frank Aune, Bartosz Belter, Wojbor Bogacki, Tim Chown, Lorenzo Colitti, Łukasz Dolata, Athanassios Duitsis, Jérôme Durand, Olivier Festor, Bartosz Gajda, Ioannis Kappas, Olav Kvittem, Roman Lapacz, Simon Leinen, Athanassios Liakopoulos, Janos Mohacsi, Wiktor Procyk, Fulvio Risso, Christian Schild, André Stolze, Robert Szuman, Bernard Tuy

* Type: P - Prototype, R - Report, D - Demonstrator, O - Other

** Security Class: PU- Public, PP – Restricted to other programme participants (including the Commission), RE – Restricted to a group defined by the consortium (including the Commission), CO – Confidential, only for members of the consortium (including the Commission)

Abstract:

Interim document reporting on any specific development carried out and performed tests of tools and management applications.

Keywords: network management, network monitoring, development


Table of Contents

1. INTRODUCTION................................................................................................................................................... 3

2. TOOLS LIST........................................................................................................................................................... 3

3. PROFORMA ........................................................................................................................................................... 5

4. DETAILED DESCRIPTION OF TOOLS DEVELOPMENT AND TESTS ..................................................... 7 Analyzer ..................................................................................................................................................................................7 Argus .......................................................................................................................................................................................9 ASpath-tree............................................................................................................................................................................12 CoralReef...............................................................................................................................................................................14 Cricket ...................................................................................................................................................................................16 Ethereal..................................................................................................................................................................................17 Flow-tools..............................................................................................................................................................................18 JOIN-TV................................................................................................................................................................................19 IPFlow ...................................................................................................................................................................................23 IPm ........................................................................................................................................................................................24 IPv6 Lan Dynamic Topology Discovery ...............................................................................................................................25 IPv6 Looking Glass ...............................................................................................................................................................30 IPv6 Management Gateway*.................................................................................................................................................32 IPv6 support for Netflow v9 in IOS.......................................................................................................................................41 Mping ....................................................................................................................................................................................42 MRTG ...................................................................................................................................................................................46 Multicast Beacon ...................................................................................................................................................................48 Nagios....................................................................................................................................................................................51 Netflow/IPFIX.......................................................................................................................................................................53 Net-snmp ...............................................................................................................................................................................54 Network Weathermap............................................................................................................................................................58 PCHAR..................................................................................................................................................................................60 Polyphemus ...........................................................................................................................................................................63 RANCID................................................................................................................................................................................64 RIPE NCC Test Traffic Server with IPv6 support .................................................................................................................66 SNMP_Session.pm ................................................................................................................................................................71 Westhawk's Java SNMP stack ...............................................................................................................................................72

5. SUMMARY ........................................................................................................................................................... 75

6. BIBLIOGRAPHY ................................................................................................................................................. 76

2


1. Introduction

There are many tools supporting network monitoring and management today. They differ from each other in their complication level and licensing type. Various tools from simple scripts to sophisticated management platforms were proposed and described in deliverable D6.2.1.

Concise descriptions of many tools included in deliverable D6.2.1 presenting their functionality and destination are very helpful for WP6 participants interested in network monitoring. Those descriptions make it easier to find an appropriate tool to match very specific expectations and requirements. Deployment of every tool is preceded by tests which evaluate their usefulness according to the network operators’ requirements.

For every tool mentioned in deliverable D6.2.1 there is a 6NET participant responsible for their continuous deployment in the IPv6 network. They also perform tests to check the new functionality of the tool in the newer versions if they are available and to estimate and check its interoperability. Some participants are the authors of the tools and perform the implementation on their own. Others add extra functionality to the tool (if the source code is available) to meet their own needs.

In deliverable D6.3.2 Implementation of tools and operational procedures, which is being edited concurrently, participants will show the real use of the tool in the IPv6 test network in the 6net environment. It will focus on the evaluation of popularity concerning testing, using and developing network tools.

This document contains descriptions of the network management tools extended with the results of tests performed by a WP6 participants. Also, some remarks concerning the implementation and descriptions of features which will be implemented in the near future are included.

2. Tools list Analyzer

Argus

ASpath-tree

CoralReef

Cricket

Ethereal

Flow-tools

JOIN-TV

IPFlow

IPm

IPv6 Lan Dynamic Topology Discovery

IPv6 Looking Glass

IPv6 Management Gateway

3


IPv6 support for Netflow v9 in IOS

MPing

MRTG

Multicast Beacon

Nagios

Netflow/IPFIX

Net-snmp

Network Weathermap

PCHAR

Polyphemus

RANCID

RIPE NCC Test Traffic Server with IPv6 support

SNMP_Session.pm

Westhawk's Java SNMP stack

4


3. Proforma The following proforma is used for the tools detailed in section 4.

Name of the tool:

Version and released date:

Available platforms:

License type:

Tool location and information links, authors:

Is source available:

Name of the considered tool

Last version that will be considered within the 6Net management framework

List of supported platforms

commercial/

shareware/

adware supported/

freeware/

GPL

URL address, list of the names and e-mail addresses of the authors

Yes/No

6NET contact: Is used in 6NET:

The name of the person and/or institution within 6Net that will be the correspondent for the tool. Yes/No

IPv6 status:

Short description, if there is support for IPv6 protocol, what has to be done for full support

Brief description of the product (brief functionality etc)

A short description of the tool

Is still under development, what is the next version number?

Information if software is still under development, if yes also next version number – minor or major, what number

Planned extensions, new features ,with dates:

DE

VE

LO

PME

NT

List of the planned extensions with dates, for example:

YYYY-MM – access through the web

YYYY-MM – multicast support

Example of running implementations publicity available

Public available URL where one can check running software, for example for MRTG and RRDBAR:

http://www.ipv6.man.poznan.pl/cgi-bin/14.cgi?cfg=tunele.cfg

Other remarks:

Information concerning software and its development which does not fit to any other field

5



Tests description

Here is a section for results of the tests which can show functionality of the tool or some important part of the functionality

Test environment description (schemes, topology, configuration) Brief description of test environment including important network configuration, host configuration, network topology, platforms and list of other software used with version numbers if needed and brief description of test scenario. Test description, results (including screenshots, frame captures), observed difficulties Description of each step of the test, test results and encountered difficulties etc.

Test summary, remarks Final test summary, remarks if the test succeed or not, what have to be done to succeed etc.

6


4. Detailed description of tools development and tests

Name of the tool: Version and released date:


License type:



Analyzer 3.0 alpha (May 29th, 2003)

Win32 BSD http://analyzer.polito.it/30alpha/

Fulvio Risso, [email protected]

Yes


Fulvio Risso [email protected], Politecnico di Torino, GARR. No

IPv6 status:

YES (Complete for packet decoding, partial for packet capture filtering)


Analyzer is an advanced network sniffing and monitoring tool with the following characteristics:

· Customisable protocol decoding engine

· Customisable views of the decoded packets

· Possibility to save captured traces in HTML

· LAN Node Discovery (to discover all the hosts on the LAN)

· LAN Node Activity monitoring (to detect which host are active on the LAN)

· Host Reachability Status (to monitor the presence and the link status of remote hosts)

· Traffic monitoring (to monitor the activity of specific hosts, specific protocols, and more)

· Traffic matrix (to monitor the activity between groups of entities)

· Possibility to run as a service (Win32 only)

· HTML statistics generations


Analyzer is still under develpment. A Final 3.0 release will be out at the beginning of the autumn


DE

VE

LO

PME

NT

Required extensions to be operational within a native IPv6 environment: None.

Proposed extensions: None.

Planned extensions: None.


None (Analyzer is mainly a standalone tool).

Other remarks:

Application domains in the 6Net management framework: Analyzer can be used mainly to monitor traffic on a local area network.

7

http://analyzer.polito.it/30alpha/

mailto:[email protected]



Tests description

Analyzer has not been tested because it has not reached a stable release.

8


Name of the tool:



License type:



Argus 3.2 relased 06/26/2003 Linux

freeware/

GPL(?) nothing mentioned

http://argus.tcp4me.com, Jeff Weisberg, [email protected] Yes


André Stolze JOIN, [email protected] Yes

IPv6 status:

IPv6 is supported as far as it is supported in perl or the used probes.


Argus is a system and network monitoring application which includes

IPv6 support since version 3.2. It will monitor nearly anything you ask it to

(TCP and UDP applications, IP connectivity, SNMP OIDS, etc). It comes with a

nice and clean, easy to view web interface that will keep both the managers and the technicians happy. Argus contains builtin alert notification via email and

pager (qpage) but is easily extendible to use any other program like i. e.

winpopup etc. It will automatically escalate alerts until they are acknoledged

by resending the alert at different intervals while optionally switching to

other methods of notification or other recipients.

Due to the fact that most of the testing modules are written in perl IPv6

functionality is included in most of them.


The tool is under development actual release is 3.2.1 and was released at 07/15/2003


DE

VE

LO

PME

NT

� Distributed Argus Redundancy Protocol

� internationalisation, i18n.

� WAP / WML / XHTML-MP

� SNMPv3

� improved https support


https://www.join.uni-muenster.de/cgi-bin/arguscgi?func=login user:6net password:<any>

Other remarks:

NONE

9

https://www.join.uni-muenster.de/cgi-bin/arguscgi?func=login


Tests description

Test environment description (schemes, topology, configuration) JOIN uses the tool to monitor its Routers and Servers mainly systems within the 6WiN. So we cannot give much configs. The Argus-config for the public part is as following example: Group "6Net" { graph: yes title: Demo for 6Net drawgrid: yes xlabel: Time ylabel: Time to reply Host "ftp6" { details: FTP-Server (v4/v6) title: FTP-Verfügbarkeit (v4/v6) Service Ping { hostname: 128.176.191.21 frequency: 180 details: Ping-Test title: Ping-Statistik (IPv4) } Service Ping { hostname: 2001:638:500:101:201:2ff:fedd:5056 frequency: 180 details: Ping-Test (IPv6) title: Ping-Statistik (IPv6) } Service TCP/FTP { hostname: 128.176.191.21 frequency: 180 details: FTP title: FTP-Availiability (IPv4) } Service TCP/FTP { hostname: 2001:638:500:101:201:2ff:fedd:5056 frequency: 180 details: FTP (IPv6) title: FTP-Availiability (IPv6) } } Host "6-WiN - 6Net Link" { details: 6-WiN - 6Net Link title: 6-WiN - 6Net Link Service Ping { hostname: 2001:798:14:200::2 frequency: 120 details: 6WiN towards 6Net title: 6WiN -> 6Net } Service Ping { hostname: 2001:798:14:200::1 frequency: 120 details: 6Net towards 6WiN title: 6Net -> 6WiN } } }

Test description, results (including screenshots, frame captures), observed difficulties JOIN just configured the tool and waited for output. It should be mentioned, that sometimes a longer or shorter frequency is required.

10


Test summary, remarks

The tests were very successful, so that we use the tool in daily work.

11


Name of the tool: Version andreleased date:


License type:

Tool location andinformation links, authors:


ASpath-tree 4.2

Solaris, FreeBSD, Linux with Perl

restricted, must beregistered tobe able todownload interpreter

http://carmen.ipv6.tilab.com/ipv6/tools/ASpath-tree/; TILAB, Italy

Yes


Isabelle ASTIC, Olivier FESTOR (INRIA), Ioannis Kappas (DANTE); János Mohácsi (NIIF/HUNGARNET) Yes

IPv6 status:

It is developed for monitor the 6Bone BGP routing table. In version 4.x it is extended to monitor IPv6 RIRaddress space as well. The BGP routing table is obtained via RSH (with some patch via telnet) from the router. If this method does support IPv6, then ASpath-tree is completely IPv6 compliant.


ASpath-tree is a tool to perform IPv6 network operation analysis based on the snapshot of the BGP routing table on IPv6 routers running BGP. Originally Aspath-tree designed to be used by an IPv6 site involved in the experimentation of the BGP protocol inside the 6Bone network, it now supports a set of features useful withinany operational IPv6 network, which makes use of BGP. The tool currently supports Cisco/Juniper/Zebrarouters. Based on a single snapshot of the IPv6 BGP table, ASpath-tree automatically generates a set of html pages providing a graphical view of the routing paths towards the other IPv6 connected domains. Additionally it provides pages for the detection of anomalous route entries announced through BGP (invalid prefixes andunaggregated prefixes), anomalous AS numbers (i.e. reserved or private) in use and a set of summary information such as:

• The number of route entries (valid/total/suppressed/damped/history) • The number of AS in table (total, originating only, originating/transit, transit only, private and

reserved) • The number of active AS paths • The number of active BGP neighbours (i.e. announcing routing information) • An analysis of the network size, in terms of AS distances • The number of circulating prefixes (total, 6Bone pTLAs, sTLAs, 6to4, others)

Based on repeated snapshots of the IPv6 BGP table at different points in time, ASpath-tree automatically generates html pages reporting on BGP routing stability (last 24 hours) for:

• 6Bone pTLAs • RIR's assigned sTLAs

Application domains in the 6Net management framework ASpath-tree will be very useful to verify the routing tables of the backbone, to check every configuration and to have some statistics about the backbone routers.


Probably version 4.3

DE

VE

LO

PME

NT


12

http://carmen.ipv6.tilab.com/ipv6/tools/ASpath-tree/





Internal modification has done to support downloading the BGP table via telnet interface not only via RSH, This modification available in the FreeBSD port collection.


6NET backbone: (version3.3)http://6nettools.dante.net/ASpath-tree/bgp.html

CERN: (version 3.3)http://www-ipv6.cern.ch/ASpath-tree-v3_3/htdocs/bgp/bgp.html

NIIF/HUNGARNET: (version 4.2)http://6net.iif.hu/6netaspathtree/

JOIN/DFN: (version 4.1)http://www.join.uni-muenster.de/bgp/bgp.html

PSNC http://www.ipv6.man.poznan.pl/bgp/bgp.html (version 4.1)

SWITCH http://www.switch.ch/network/ipv6/bgp/ (version 3.3)

UNINETT http://drift.uninett.no/ipv6/bgp/bgp.html (version 4.1)

Other remarks:

Could be extended to support Windows environment, but no effort was put on it.

Tests description

The operator of the sites above usually looks everyday to the web pages to find out BGP routing misconfiguration. They use it for everyday operational management to check routing stability and assess IPv6 routing reachability.

Test environment description (schemes, topology, configuration) The sites above use different machines and operating systems (Solaris, Linux and FreeBSD)and the tools seems to be stable and capable of providing information about the IPv6 BGP routing. The tool is tested to collect BGP routing information from Cisco 12404, Cisco 7500 and Cisco 7200 and Cisco 3660 routing platform. Most of the sites are using RSH to transfer BGP routing table. Some test has been done to test telnet BGP routing table collection at 6NET core (by DANTE) and at NIIF/HUNGARNET. Test description, results (including screenshots, frame captures), observed difficulties The telnet interface is working successfully as displayed the above sites. The telnet interface not working properly over IPv6 some work has to be done to support IPv6 properly.

Test summary, remarks The above sites use Aspath-tree for everyday operation. If you need a quick overview of IPv6 routing at your site you can run the tool.

13

http://6nettools.dante.net/ASpath-tree/bgp.html

http://www-ipv6.cern.ch/ASpath-tree-v3_3/htdocs/bgp/bgp.html

http://6net.iif.hu/6netaspathtree/

http://www.join.uni-muenster.de/bgp/bgp.html


Name of the tool:



License type: Tool location and information links, authors: Is source

available:

CoralReef Version 3.5.1

FreeBSD,

Linux,

Solaris,

And expected to work on most Unix-like systems

http://www.caida.org/tools/measurement/coralreef Yes


[email protected] No

IPv6 status:

CoralReef was partially ported over IPv6 : the Transport Level and the application making IP-level statistiques (crl_rate) were already IPv6 enabled.

We ported over IPv6 the application which monitor flows (crl_flow) and the HTML pages generator (t2_report)


CoralReef is a set of applications to collect, analyses and report IP Traffic. It could work on passive traffic or from trace files given by tcpdump or dagtools.

Its main applications are crl_rate, crl_flow and t2_report.

Crl_rate collects IP statistiques (nb of paquets, nb of bytes, IPv4 packets, IPv6 packets.

Crl_flow distinguishes the multiple flows and make post-analysis report using t2_report. For example, it separates flows depending on the type of Transport Protocol, or on the couple (source address, destination address)

T2_report displays the output of Crl_flow through a Web interface. The user could find the output in txt format and a pie chart summarizing the proportion of each flow among the whole traffic.



DE

VE

LO

PME

NT


http://www.caida.org/dynamic/analysis/workload/sdnap/

Other remarks:

14

http://www.caida.org/tools/measurement/coralreef


http://www.caida.org/dynamic/analysis/workload/sdnap/


Tests description

We tested it in the context of a Gigabit network.

Test environment description (schemes, topology, configuration) All our systems were on FreeBSD 4.7. We used it to monitor the traffic between two systems : one generating the traffic, the second monitoring it. Test description, results (including screenshots, frame captures), observed difficulties Coralreef give interesting information when used on a 100Mbits network. But it lost a lot of packet in the case of a Giga network and thus, the information it gave were erroneous.


CoralReef is a very interesting tool to visualize the distribution of the traffic, and to see how the network is used.

15


Name of the tool:



License type:



Cricket 1.0.3 http://cricket.sourceforge.net/


6Net Contact: Ioannis Kappas (DANTE), Simon Leinen (SWITCH) Janos Mohacsi (HUNGARNET) Yes

IPv6 status:

IPv6 Support: NO


Network operators require awareness of how well their network performs. Every node in the network keeps statistics on many attributes that affect its performance. The operators would like to constantly monitor these attributes over time and keep track of their intensity.



DE

VE

LO

PME

NT

Required extensions to be operational within a native IPv6 environment The SNMP polling method can be developed further to operate over a native Ipv6 network.

Proposed extensions SNMP is the most supported protocol for accessing attributes from network nodes. It has been proposed in the context of the 6NET project to use SNMPv3 capable tools that can operate inside a native Ipv6 domain. A new polling method that interfaces with an existing Ipv6 SNMPv3 library (e.g. Net-SNMP) can be defined as an alternative to upgrading the internal SNMP polling method.


http://6net.iif.hu/cricket/grapher.cgi (HUNGARNET)

Other remarks:

Application domains in the 6Net management framework The tool can be used by anyone who wants to monitor and plot value variations of network attributes inside their management domain

16

http://cricket.sourceforge.net/

http://6net.iif.hu/cricket/grapher.cgi


Name of the tool:



License type:



Ethereal 0.9.13 List of supported platforms

GPL http://www.ethereal.com/ Yes


Simon Leinen, SWITCH Yes (by several participants)

IPv6 status:

IPv6 Support: YES


Ethereal is a packet analyser with a graphical (GTK) front-end that supports drill-down



DE

VE

LO

PME

NT

Required extensions to be operational within a native IPv6 environment None. Ethereal fully supports the basic IPv6 protocols, and all TCP- and UDP-based application protocols running over IPv6. It is widely used to develop and troubleshoot IPv6 applications and protocols.

Proposed extensions Protocols that are used or developed within 6NET could be supported with additional or improved dissectors if required.


na

Other remarks:

Application domains in the 6Net management framework Network Element troubleshooting, network fault isolation, intrusion detection

17

http://www.ethereal.com/


Name of the tool:



License type:



Flow-tools 0.66 BSD style

http://www.splintered.net/sw/flow-tools/ Yes


Ioannis Kappas (DANTE) No

IPv6 status:

IPv6 Support: No


Network equipment can account and report traffic statistics for the purpose of network management. It is normally a requirement of a network operator to keep track how much and what type of traffic the network clients send and which networks or end systems are receiving it.



DE

VE

LO

PME

NT

Required extensions to be operational within a native IPv6 environment: The tool should be upgraded to include support for Netflow export v9, which can incorporate information about IPv6 flows. It should also be upgraded to use the IPv6 stack in addition to the only one currently supported Ipv4 stack.




Other remarks:

Application domains in the 6Net management framework: Flow-tools can be deployed in any of the domains that need to collect flow information for traffic transiting their border routers.

Tests description

The flow-tool is the most popular flow collection and analyses tool in the community and provides many general features that satisfy most of network operators. It supports Netflow v5 and v8 at least; though these versions can only provide IPv4 information about flows but not IPv6. The new netflow specification (v9) is general enough and includes fields specific to IPv6; though because it is a recent standard and is not present yet on production environments cflow-tools does not support it. So it is not possible to perform any tests at this moment.

18

http://www.splintered.net/sw/flow-tools/

http://www.splintered.net/sw/flow-tools/



Name of the tool:



License type:

Tool location and informationlinks, authors:


JOIN-TV 0.6.1 Unix/Linux GPL [email protected] Yes


André Stolze ([email protected])

JOIN ([email protected]) Yes

IPv6 status:

Supports IPv6 depending on the IPv6-Support of MRTG.


JOIN-TV (JOIN Traffic Visualizer) visualizes the Traffic-flow within a network by interpreting the log-files of the MRTG-Tool written by Tobias Oetiker.


Still under development (see below)

Planned extensions, new features, with dates:

DEVELOPMENT

- Detailed documentation

- Localization

- Extended support for background maps

- Finetuning of display output and modular style

- Traffic history


http://www.6win.de/mrtg/index_neu.php?lang=en

http://www.6win.de/mrtg/index_abs.php?lang=en

Other remarks:

Tests description

Test environment description (schemes, topology, configuration) Software: All tests were performed on the JOIN webserver which is currently a dual Intel-Xeon 2.4 GHz 1HE server-system running Mandrake Linux release 9.0 (dolphin) for i586. For security reason the MRTG-tool(2.9.25) is running on a different system, an AIX-System gathering data via IPv4 (this special machine does not support IPv6). Data transfer is done by a cron job executing a simple “scp” command. The interpretation of data is done by the php-module (v4.3.1) for the apache-webserver (2.0.45) with external GD support including full/maximal freetype2-support.

19


http://www.6win.de/mrtg/index_neu.php?lang=en


Network: JOIN-TV was developed for the 6WiN (http://www.6win.de) – a native IPv6 backbone in Germany – and the tunnel endpoints of the connected members of DFN. Therefore all tests of the tool were done within the 6WiN. All monitored interfaces in the tests are located on Cisco 7206 routers equipped with currently IOS 12.2(15)T. For this reason there are two values for the inner-6WiN-links and only one for customer endpoints.

Test description, results (including screenshots, frame captures), observed difficulties Test results can be viewed at the WWW-Addresses mentioned above. Clicking on a link opens a new window containing the MRTG graphs for the selected link. Figures 1 and 2 show example pictures of JOIN-TV.

20


Figure 1 : 6WiN traffic overview (bandwidth usage in percent)

21


Figure 2 : 6WiN traffic overview (absolute bandwidth usage)

22


Name of the tool:



License type:



IPFlow 0.49.7pre3 http://www.rrt.cr-picardie.fr/~fillot/nf6/ No


Simon Leinen, SWITCH Yes (SURFnet, HUNGARNET)

IPv6 status:

IPv6 Support: YES


IPFlow is a collector for Netflow version v1, v5, v6, v7, v8 and v9. It supports logging flow data to disk, data aggregation according to configuration, port scan detection, storage of aggregated data in RRDtool, and graphical display of flow statistics. The author is Christophe Fillot.



DE

VE

LO

PME

NT

Proposed extensions Support for the IPFIX protocol as it becomes standardized in the IETF.


Other remarks:

Application domains in the 6Net management framework See NetflowIpFix

23

http://www.rrt.cr-picardie.fr/~fillot/nf6/

http://www.rrt.cr-picardie.fr/~fillot/nf6/

http://6nettools.dante.net/cgi-bin/moin/moin.cgi/NetflowIpFix


Name of the tool:



License type:



IPm IPm2.2 List of supported platforms

commercialhttp://www.efficientip.com



Jérôme DURAND No

IPv6 status:

IPm supports IPv6 DNS management


This tool provides facilities to manage:

Naming

Registry (addressing)

DHCP

Authority management (for network administrtors rights)


No information


DE

VE

LO

PME

NT

No information

There should be support for IPv6 registry soon (next IPv6 step)


No running implementation publicly available

Other remarks:

No remarks at this stage

Tests description

No tests performed yet. Should be done soon.

Only installation and global features were seen (nothing in detail)

24


Name of the tool:



License type:



IPv6 Lan Dynamic Topology Discovery V1.0 FreeBSD freeware [email protected] Yes


Isabelle ASTIC, INRIA-LORIA

Olivier FESTOR, INRIA-LORIA No

IPv6 status:

Developped for IPv6 LANs.


This service dynamically discovers the IPv6 LAN topology, thanks to its hierarchical architecture based on a Local Agent (LA) working on each IPv6 link, and a Global Agent(GA) collecting and correlating all the data received from the LAs. A Visualisation agent (VA) is connected to the GA and periodically display the view of the topology.


This service is still under developpement. Some tests need to be perform yet.


DE

VE

LO

PME

NT

See test Summary and remarks


Other remarks:

Development description

The current dynamic topology discovery for IPv4 networks rely on ICMP and SNMP. To discover all the nodes of a sub-network, they use exhaustive pings in order to scan all possible addresses and verify that they are used. To discover the multiple sub-networks, SNMP is used to learn the routing table and the « Next Hops » on each route. As mostly SNMP architecture is not available (SNMP over IPv6 is quite rare and the MIB able to manage IPv6 networks too), and as scan all the possible addresses on a sub-network is much more complex now because of the length of the IPv6 addresses, a new LAN dynamic topology services should be design.

25


The study of the IPv4 services, of the available IPv6 protocols to find out a IPv6 network topology, of the specificity of IPv6 (addressing architecture), induce to propose a hierarchical architecture based on several Local Agents (LA), one agent implemented on each local link, collecting the topology of this link, and a unique Global Agent (GA)which receive the information collected by each LA and correlate them in order to establish the real topology of the IPv6 LAN (see Figure 1 below).

The LA algorithm relies on ICMPv6 and Neighbor Discovery protocol. It first send an Echo Request to all the nodes and collect more information about this node, using the neighbor discovery.

The GA algorithm collects all the information from the LA. It correlates them in order to discover interfaces detected by 2 different LAs but owned by the same node. It relies each LA by a traceroute to determine the main routes. So the topology discovered is a N3 topology.

Host

Local Agent

Host Host Host

Local Agent

Local Agent

Visualisation Agent

Link topology discovery

Link topology discovery

Visualisation Agent

Visualisation Part

Gather information

Host

Figure 3 : IPv6 Lan Topology discovery architecture Way of the data

26


The Visualisation Agent connects itself to the GA. It then receives the topology discovered by the GA and refreshes it periodically. Multiple VA could be connected to a GA. Figure 2 give an exemple of topology as displayed by the VA :

Figure 4 : Exemple of a topology view displayed by the Visualisation Agent

The protocol used between the different agents is UDPv6. The visualisation of the IPv6 LAN topology is made using QT.

This service was implemented in the C language, on a FreeBSD platform (see description of the tests below).

Tests description

The tests made were made in the context of the development of the service.

Test environment description (schemes, topology, configuration) They were made on our IPv6 testbed platform. This platform is linked to the main IPv4 LORIA network by a switch on our testbed platform and a dual stack CISCO router : the LORIA IPv6 gateway. All our systems are under FreeBSD 4.7. A DNS server is available on our file server (Rork). Rork is defined with trunked VLANs which allow it to be joined from every link of our platform. See figure 1 below for a schematic description of our platform.

27


Xl0

Xl0 Xl0

Xl0

Xl0 Xl2

Xl1

Xl2Xl1

Xl1

Xl1

Aria

2001:660:301:40::/64

Router

2001:660:301:32::/64

Rork1 Loria Ipv6 gateway

2001:660:301:33::/64

2001:660:301:34::/64

Garfield

Thorgal

2001:660:301:35::/64

Asterix Rork2

Treize

Sha Rork3

2001:660:301:37::/64

Rork4

2001 :660 :301 :38 ::/64

Rork5

host

Legend :

Crossed wire

Rork6

canardo betelgeuse

Figure 5 : IPv6 LORIA Testbed platform

The Local Agents are set on Thorgal, Treize, Asterix and Aria. The LA sest on Thorgal manages the link connected to its xl1 interface. On Treize, the LA manages the link connected to its xl2 interface. On Asterix, two LAs are set, one for each interface. And at least, the LA sets on Aria manages the link connected to its xl0 interface. The Global Agent is implemented on Aria too.

Test description, results (including screenshots, frame captures), observed difficulties

The results of each Local Agent are summarized into the next table (N means Node, R means Router) (Figure 2):

Local agent Sub-network Name IPv6 link-local address IPv6 global address Link layer address Type

Aria xl0 2001:660:301:32::/64 Rork1 Fe80::201:2ff:fee3:6019 N

Canardo Fe80:2b0:d0ff:fe3f:8260 2001:660:301:32:2b0:d0ff:fe3f:8260 00:b0:d0:3f:82:60 N

Betelgeuse Fe80::260:8ff:fe50:cbe9 2001:660:301:32:260:8ff:fe50:cbe9 00:60:08:50:cb:e9 N

Loria GW Fe80::230:b6ff:fe51:d41c 00:30:b6:51:d4:1c R

Aria Fe80::201:2ff:fee3:608a 2001:660:301:32:201:2ff:fee3:608a 00:01:02:e3:60:8a R

Aria xl1 2001:660:301:33::/64 Rork2 Fe80::201:2ff:fee3:6019 N

Asterix Fe80::201:2ff:fee3:5fcc 2001:660:301:33:201:2ff:fee3:5fcc 00:01:02:e3:5f:cc R

Aria Fe80::201:2ff:fee3:605d 2001:660:301:33:201:2ff:fee3:605d 00:01:02:e3:60:5d R

Asterix xl1 2001:660:301:34::/64 Rork6 Fe80::201:2ff:fee3:6019 N

Treize Fe80::201:2ff:fee3:6013 2001:660:301:34:201:2ff:fee3:6013 00:01:02:e3:60:13 R

Thorgal Fe80::210:4bff:fecd:e299 2001:660:301:34:210:4bff:fecd:e299 00:10:4b:cd:e2:99 R

Asterix Fe80::201:2ff:fee3:6015 2001:660:301:34:201:2ff:fee3:6015 00:01:02:e3:60:15 R

Thorgal xl1 2001:660:301:35::/64 Rork3 Fe80::201:2ff:fee3:6019 N

28


Sha Fe80::2c0:4fff:febb:af7a 2001:660:301:35:2c0:4fff:febb:af7a 00:c0:4f:bb:af:7a N

Thorgal Fe80::2c0:4fff:fe67:6b32 2001:660:301:35:2c0:4fff:fe67:6b32 00:c0:4f:67:6b:32 R

Treize xl2 2001:660:301:37::/64 Rork4 Fe80::201:2ff:fee3:6019 N

Treize Fe80::210:4bff:fecd:defd 2001:660:301:37:210:4bff:fecd:defd 00:10:4b:cd:de:fd R

Garfield xl2 2001:660:301:38::/64 Rork5 Fe80::201:2ff:fee3:6019 N

Garfield Fe80::201:2ff:fee3:6005 2001:660:301:38:201:2ff:fee3:6005 00:01:02:e3:60:05 R

Treize xl1 2001:660:301:40::/64 Garfield Fe80::201:2ff:fee3:6089 2001:660:301:40:201:2ff:fee3:6089 00:01:02:e3:60:89 R

Treize Fe80::250:daff:fedc:8ebb 2001:660:301:40:250:daff:fedc:8ebb 00:50:da:dc:8e:bb R

Figure 6 : Local link discovery algorithm results

The main encountered difficulties were to find out the IPv6 global address of each system. The algorithm is based on the Neighbor Discovery and the ICMPv6 Protocol. These 2 algorithms assure to discover the link-local address and the physical one. If the IPv6 global address of an equipment is not stateless auto-configured, it is sometimes impossible to find out its IPv6 global address. That is the case of the LORIA IPv6 Gateway, which IPv6 global address is missing, and which is manually configured. The second difficulty is to establish that two interfaces are owned by the same system. Using DNS is most of the time sufficient but sometimes more information is needed, as in the case of our file server Rork whom interface owns each a different name. Test summary, remarks Our algorithm gives all the information that it could discovered. Because of the lack of the MIB II for IPv6 networks and, in the case of native networks, of the lack of SNMP over IPv6, the difficulties encountered could not be solved. When the MIB II and SNMP over IPv6 will be widely implemented, then, it will be possible to discover more information on all systems but only the use of protocol like the Matt Crawford one, which is an extension of ICMPv6 (IPv6 Node Information Queries, draft-ietf-ipngwg-name-lookups-09;txt), could help to discover a complete topology.

29


Name of the tool:



License type:



IPv6 Looking Glass - - freewarew6.loria.fr [email protected]

Yes


Isabelle ASTIC

Olivier FESTOR Yes

IPv6 status:

Full IPv6 support


The Looking Glass is composed of a Web interface generated by a CGI file. This Web interface let the user choose several requests to be sent to CISCO routers. Each request corresponds to a CISCO CLI command. These commands are executed on CISCO routers via a telnet connexion over IPv6.



DE

VE

LO

PME

NT

Extension could be made by extended the Looking Glass to other types of routers.


w6.loria.fr

Other remarks:

[email protected]


30


The Looking Glass existed for IPv4 networks. Our development consisted on porting it on IPv6 (modification of the Net::telnet::Cisco module) and to define the Web page with the most interesting commands to be sent (see a copy of the HTML page below) :

Tests description

This IPv6 version of Looking Glass was mainly used to manage the IPv6 Pilote of Renater II.

It is useful to verify the behaviour of the routing protocols and the traffic on the network.

31


Name of the tool:

Version and

released date:


License type:



IPv6 Management Gateway*

1.0

2003.06.06 Linux GPL

http://www.ipv6.man.poznan.pl Bartosz Belter: [email protected] Wojbor Bogacki: [email protected] Łukasz Dolata: [email protected] Marcin Kamiński: [email protected]

Yes


[email protected] Yes

IPv6 status:

Support SNMP(without MIB conversion), and ICMP.


The main purpose of the developed IPv6 Management Gateway is to enable the existing IPv4 network management platforms to monitor, configure and manage the native IPv6 network. The IPv6 Management Gateway translates SNMP and ICMP protocol messages between IPv4 and IPv6 networks.


Tool is still under development.


DE

VE

LO

PME

NT

Support for certificate authorization.

Autoconfiguration.

Module for simple TCP transition (for example telnet, ssh and others).

Access levels for users (read only, read-write) in TT Manager.


http://chives.man.poznan.pl

Other remarks: * Old name of IPv6 Management Gateway 1.0 was SNMP Transition Tool 2.0

Tests description Test environment description (schemes, topology, configuration)

Topology:

32

http://www.ipv6.man.poznan.pl/






http://chives.man.poznan.pl/


IPv4 network IPv6 network

Data Generalbirch.man.poznan.pl10.0.7.3

baobab,man.poznan.pl10.0.7.2

3Com

baobab.man.poznan.pl2001:808:0:1:204:75ff:fec7:c277

3Com

fernek.man.poznan.pl2001:808:0:1:210:4bff:feb8:4da9

fernek,man.poznan.pl10.0.7.4

boletus,man.poznan.pl10.0.7.1

3Co m

birch.man.poznan.pl2001:808:0:1:200:1ff:fe00:c43f

3Com

boletus.man.poznan.pl2001:808:0:1:204:76ff:fee9:59d9

fern.man.poznan.pl10.0.7.5

2001:808:0:1:204:75ff:fec7:5a4b

Server proxy

IPv4 network IPv6 network

Data Generalbirch.man.poznan.pl10.0.7.3

baobab,man.poznan.pl10.0.7.2

3Com

baobab.man.poznan.pl2001:808:0:1:204:75ff:fec7:c277

3Com

fernek.man.poznan.pl2001:808:0:1:210:4bff:feb8:4da9

fernek,man.poznan.pl10.0.7.4

boletus,man.poznan.pl10.0.7.1

3Co m

birch.man.poznan.pl2001:808:0:1:200:1ff:fe00:c43f

3Co m

boletus.man.poznan.pl2001:808:0:1:204:76ff:fee9:59d9

fern.man.poznan.pl10.0.7.5

2001:808:0:1:204:75ff:fec7:5a4b

Server proxy

Configuration:

Computer name: boletus

Processor: Intel Celeron 1.0 GHz

Memory: 256 MB

Network card: 3Com EtherLink XL 10/100 Mbit

Pla form: Linux Red Hat 8.0 t

t

IPv4 address: 10.0.7.1

IPv6 address: 2001:808:0:1:204:76ff:fee9:59d9

Computer name: baobab


Memory: 256 MB


Pla form: Linux Mandrake 9.0 IPv4 address: 10.0.7.2

IPv6 address: 2001:808:0:1:204:75ff:fec7:c277

33


Computer name: birch

Processor: Intel Celeron 700 MHz

Memory: 256 MB


Pla form: Linux Suze 8.0 t

t

t

IPv4 address: 10.0.7.3

IPv6 address: 2001:808:0:1:200:01ff:fe00:c43f

Computer name: fernek


Memory: 256 MB


Pla form: Linux PLD 1.0 IPv4 address: 10.0.7.4

IPv6 address: 2001:808:0:1:210:4bff:feb8:4da9

Computer name: fern


Memory: 256 MB


Pla form: Linux PLD 1.0 IPv4 address: 10.0.7.5

IPv6 address: 2001:808:0:1:204:75ff:fec7:5a4b

The computers was connected by HUB 3Com – LinkBuilder TP/8, and the NET-SNMP 5.0.6 was installed on every machine.

Address translation table. 10.0.7.1 2001:808:0:1::11

10.0.7.2 2001:808:0:1::12

10.0.7.3 2001:808:0:1::13

10.0.7.4 2001:808:0:1::14

10.0.7.5 2001:808:0:1::15

2001:808:0:1:204:76ff:fee9:59d9 10.0.7.11

2001:808:0:1:204:75ff:fec7:c277 10.0.7.12

34


2001:808:0:1:200:01ff:fe00:c43f 10.0.7.13 2001:808:0:1:210:4bff:feb8:4da9 10.0.7.14 2001:808:0:1:204:75ff:fec7:5a4b 10.0.7.15

Test description, results (including screenshots, frame captures), observed difficulties Test 1:

Description:

Every station performs a snmpwalk towards all agents simultaneous using the IPv6 Management Gateway.

The IPv6 Management Gateway translates SNMP packets between IPv4 and IPv6 networks in both directions.

Results:

[bart@birch bart]$ snmpwalk –c public 2001:808:0:1::14

SNMPv2-MIB::sysDescr.0 = STRING: Linux fernek 2.2.22 #1 Sun Nov 17 10:56:45 UTC 2002 i586

SNMPv2-MIB::sysObjectID.0: NET-SNMP-MIB::netSnmpAgentOIDs.10

SNMPv2-MIB::sysUpTime.0 = Timeticks: 0:08:51.31

SNMPv2-MIB::sysContact.0 = STRING: Me <[email protected]>

SNMPv2-MIB::sysName.0 = STRING: fernek

SNMPv2-MIB::sysLocation.0 = STRING: Right here, fight now.

SNMPv2-MIB::sysORLastChange.0 = Timeticks: 0:00:01.42

SNMPv2-MIB::sysORID.1 = OID: IF-MIB::ifMIB

SNMPv2-MIB::sysORID.2 = OID: SNMPv2-MIB::snmpMIB

SNMPv2-MIB::sysORID.3 = OID: TCP-MIB::tcpMIB

SNMPv2-MIB::sysORID.4 = OID: IP-MIB::ip

SNMPv2-MIB::sysORID.5 = OID: UDP-MIB::udpMIB

SNMPv2-MIB::sysORID.6 = OID:SNMP-VIEW-BASED-ACM-MIB::vacmBasicGroup

SNMPv2-MIB::sysORID.7 = OID: SNMP-FRAMEWORK-MIB::snmpFrameworkMIBCompliance

SNMPv2-MIB::sysORID.8 = OID: SNMP-MPD-MIB::snmpMPDCompliance

SNMPv2-MIB::sysORID.9 = OID: SNMP-USER-SM-MIB::usmMIBCompliance

SNMPv2-MIB::sysORDescr.1 = STRING: The MIB module t describe generic objects for Network interface sub-layers

SNMPv2-MIB::sysORDescr.2 = STRING: The MIB module for SNMPv2 entities

SNMPv2-MIB::sysORDescr.3 = STRING: The MIB module for managing TCP implementations

[bart@birch bart]$ snmpwalk –c public 10.0.7.14

SNMPv2-MIB::sysDescr.0 = STRING: Linux fernek 2.2.22 #1 Sun Nov 17 10:56:45 UTC 2002 i586

SNMPv2-MIB::sysObjectID.0: NET-SNMP-MIB::netSnmpAgentOIDs.10

SNMPv2-MIB::sysUpTime.0 = Timeticks: 0:08:49.49

35


SNMPv2-MIB::sysContact.0 = STRING: Me <[email protected]>

SNMPv2-MIB::sysName.0 = STRING: fernek

SNMPv2-MIB::sysLocation.0 = STRING: Right here, fight now.

SNMPv2-MIB::sysORLastChange.0 = Timeticks: 0:00:00.37

SNMPv2-MIB::sysORID.1 = OID: IF-MIB::ifMIB

SNMPv2-MIB::sysORID.2 = OID: SNMPv2-MIB::snmpMIB

SNMPv2-MIB::sysORID.3 = OID: TCP-MIB::tcpMIB

SNMPv2-MIB::sysORID.4 = OID: IP-MIB::ip

SNMPv2-MIB::sysORID.5 = OID: UDP-MIB::udpMIB

SNMPv2-MIB::sysORID.6 = OID:SNMP-VIEW-BASED-ACM-MIB::vacmBasicGroup

SNMPv2-MIB::sysORID.7 = OID: SNMP-FRAMEWORK-MIB::snmpFrameworkMIBCompliance

SNMPv2-MIB::sysORID.8 = OID: SNMP-MPD-MIB::snmpMPDCompliance

SNMPv2-MIB::sysORID.9 = OID: SNMP-USER-SM-MIB::usmMIBCompliance

SNMPv2-MIB::sysORDescr.1 = STRING: The MIB module t describe generic objects for Network interface sub-layers

SNMPv2-MIB::sysORDescr.2 = STRING: The MIB module for SNMPv2 entities

SNMPv2-MIB::sysORDescr.3 = STRING: The MIB module for managing TCP implementations

Observed difficulties:

None.

Test 2:

Description:

The Boletus station with HP OpenView on the IPv4 network sends an SNMP request to the Baobab and the Fernek stations on the IPv6 network. They answer to the Boletus station. All SNMP packets are translated by the IPv6 Management Gateway installed on the Fern station.

Screenshots:

36



None.

Test 3:

Description:

The Birch station with MG-SOFT on the IPv4 network send SNMP request to the Baobab and the Fernek stations on the IPv6 network. They answer to the Birch. All SNMP packets are translated by the IPv6 Management Gateway installed on the Fern station.

Screenshot:

37



None.

Test 4:

Description:

The Birch station with SolarWinds on the IPv4 network sends the SNMP request the Baobab and the Fernek stations on the IPv6 network. They answer to the Birch. All SNMP packets are translated by the IPv6 Management Gateway installed on the Fern station.

Screenshot:

38



None. Test 5:

Description:

The Birch station with PSNC Muvi MIB-Browser on the IPv4 network sends an SNMP request to the Baobab and the Fernek stations on the IPv6 network. They answer to the Birch. All SNMP packets are translated by the IPv6 Management Gateway installed on the Fern station.

A similar test is executed on the other side, from IPv6 to IPv4 network.

Screenshots:

39



None.

Test 6:

Description

The Baobab station on the IPv4 network sends the ICMP ping message to the Boletus station on the IPv6 network. It answers to the Baobab. ICMP packets are translated by the IPv6 Management Gateway installed on the Fern station.

Results:

The Baobab station gets correct ICMP ping message from the Boletus station. The IPv6 Management Gateway on the Fern station is transparent either the Baobab station or the Boletus station.


None.


The IPv6 Management Gateway translates correctly SNMP and ICMP messages. Tests confirmed the compatibility of the IPv6 Management Gateway with other producers software. It’s also transparent for this software. No problems occurred during tests.

40


Name of the tool:



License type:



IPv6 support for Netflow v9 in IOS

lrevarde-nf6 EFT March-June 2003

C3640/ 7200/ 7500

commercial, currently distributed on a restricted basis as an EFT (Engineering Field Trial)

URL address, list of the names and e-mail addresses of the authors

No


Simon Leinen, SWITCH Yes (SWITCH, RENATER, SURFnet, DANTE)

IPv6 status:

IPv6 Support: Yes


The metering/exporting side of Netflow v9 has been implemented in Cisco IOS. Currently, 3640 and 7200/7500 routers are supported.



DE

VE

LO

PME

NT

Required extensions to be operational within a native IPv6 environment Transport of flow information over IPv6.

Proposed extensions Support of the emerging IPFIX standard.


Other remarks:

Application domains in the 6NET management framework See NetflowIpFix

41

http://6nettools.dante.net/cgi-bin/moin/moin.cgi/NetflowIpFix


Name of the tool: Version and released date:

Available platforms: License type:



Mping 2.0 RC1

Tested on:

GNU/Linux

FreeBSD

NetBSD

GPL-like

URL: http://mping.uninett.no

Email:

[email protected]

Yes


Olav Kvittem – [email protected]

Frank Aune - [email protected] Yes

IPv6 status:

Version 2.0 supports IPv6 but is currently in beta testing.


The Mping service consists of two parts: The Mping client, written in C, and the web interface extension, written in PERL. Unless otherwise specified, in this document when referring to Mping we are referring to the Mping C-client

Mping is a tool for measuring round-trip delay and packet loss, using the ICMP echo feature, in a TCP/IP based network. Multiple hosts - up to 500, both IPv4 and IPv6 at the same time - can be pinged in a round-robin order. At runtime, the user can set the wait time between each packet sent, number of packets sent and the size of the packets. For each host specified, information about packet loss and minimum/average/maximum response time is displayed. Mping can also display the collected data as median, cube-sum, standard deviation or 10/50/90-percentile at the users request.

Several techniques are implemented into the Mping service, to make sure that the collected data is “statistically” correct:

Mping by default do not send more than 10 ICMP packets per second, thus measured data is independent from the time of measuring.

Mping do not send all ICMP-packets to one 'Gateway' at the same time, rather Mping tries to spread it out in a Round-Robin fashion, thus avoids temporary network characteristics.

Mping starts the pingsweeps at asynchronous intervals. We use a Poisson-distribution, thus avoiding periodic network variance.

Technique 1 and 2 are Mping C-client features, while 3 is implemented in the PERL web interface extension.

A web interface is used for browsing the collected data and for generating reports, graphs and traceroutes for the different hosts we measure. The PERL code is modularized and easily extended to suite other needs. As an example, the language support is modularized and thus adding support for new languages is very easy.

42

http://mping.uninett.no/






Yes, the next version planned is v2.0.


DEVELOPMENT

2003-08 – Mping RC1 revision.

2003-08 – Store historical traceroute for each host monitored, so we can see what route the packets travelled for any given date.

2003-08 – Improve graph zooming / scaling / navigation in the web interface.


http://mping.uninett.no

Other remarks:

Tests description

Test environment description (schemes, topology, configuration) SYSTEM CONFIGURATION: Mping 1.3 (IPv4 only) has been tested on the following platforms successfully:

1.FreeBSD 4.8 Production / 5.1 New Technology 2.NetBSD 1.6 STABLE 3.GNU/Linux 2.2.X and 2.4.X

Mping 2.0 (both IPv4 and IPv6) has been tested on the following platforms successfully:

GNU/Linux >= 2.4.18* (See note below) NetBSD 1.6 STABLE (only IPv4 support tested)** (See note below) FreeBSD 4.8 Production / 5.1 New Technology (only IPv4 support tested)** (See note below)

* Mping relies on the ICMP IPv6 packet checksum feature of the Linux 2.4.X kernel series, however we have been seeing strange issues on 2.4.X kernels below 2.4.18. Even on some 2.4.18 kernels, we have observed this issue. For RedHat at least 2.4.18-27.7 and above is working as expected. If IPv6 is going to be used, we suggest using no kernel older than this. ** No suitable test environment for NetBSD and FreeBSD with regards to IPv6 support were available at the time of writing. However, we see no reason why this should not work. Since Mping uses raw sockets to send the ICMP packets, the client needs to be setuid and owned by root (at least have permissions to access raw sockets).

43



The traceroute feature for the PERL web interface, requires iputils => 200209, due to a bug in traceroute6 in older iputils-packages. The PERL installation for the web interface, requires either the GIFGraph (or GDGraph) module and gnuplot, to be able to produce graphs from the collected data. The Mping C-client is packaged in a separate tar.gz file, and support normal *nix Makefile operations. The web interface extension has its own relatively easy installation procedure described in the README file included in the package.

NETWORK CONFIGURATION: IPv6 over IPv4 tunnelling has been used in the testing phase, so we have yet to determine if encapsulation of packets this way has had any impact on the data collected or performance observed.

44


Test description, results (including screenshots, frame captures), observed difficulties Currently Mping 2.0 is tested against several groups of hosts, including the 6Net router backbone, a mixed IPv4 / IPv6 group and a group with other hosts residing on 6Net. (See http://mping.uninett.no ) EXTERNAL DIFFICULTIES: As noted above, on Linux kernel <2.4.18 we've observed issues with IPv6 packet checksum generation, resulting in Mping doing pretty much nothing – since it believes the packets received are bad due to kernel checksum generation was flawed. This can easily be rectified by using >=2.4.18-27.7 (RedHat patch) or some other up-to-date patched Linux kernel >=2.4.18. INTERNAL MPING DESIGN DIFFICULTIES: When listening to two sockets at once, to be able to support both IPv4 and IPv6 in parallel, we also had to make sure they did not block each other – and thus the Mping design is somewhat different than a normal ping client found on most *nix OSes. The other major “difficulty” encountered, was the decision on how to design Mping when a DNS hostname resolves more than one address, and perhaps it resolves both an IPv4 and IPv6 address. We have implemented an optional -4 and -6 switch set at runtime, so the user can specify the desired protocol if more than one is available, and have yet to see the need for a more sophisticated behaviour. If no protocol is specified, Mping just takes the first address it resolves. Test summary, remarks All tests so far have been successful, and the Mping client is working as expected. The web interface is currently undergoing changes, but the data collection is unaffected by this. Since the Mping client is written in C and the web interface is based on PERL, they do not share any mutual code at all (the web interface is simply using the data output from the Mping client in its presentation). Remark: How efficiently a network stack is able to handle a high load, is very different from platform to plattfom. Generally, a machine under heavy load is unsuited to perform accurate network measurements. Ideally, one would want one machine to only collect the data, and use another machine to manipulate this data and host the web interface etc. The machine running our Mping service is also used for other tasks, and how this affects the statistics is something we are investigating.

45



Name of the tool:




available:

MRTG 2.10.0pre3

04/06/2003

Linux, FreeBSD, Solaris, Win32

GPL http://people.ee.ethz.ch/~oetiker/webtools/mrtg/ Yes


Roma Tre Computer Networks research group Yes (IPv4 only)

IPv6 status:

Supports SNMP over IPv6 (on Linux, probably also on FreeBSD, but probably not on Win32).

Does not yet support IPv6 MIBs for querying/displaying IPv6 addresses, etc. Brief description of the product (brief functionality etc)

The Multi Router Traffic Grapher (MRTG) is a tool to monitor the traffic load on network-links. MRTG generates HTML pages containing graphical images which provide a LIVE visual representation of this traffic.


2.10.0


DE

VE

LO

PME

NT

(unknown date) – IPv6 MIB support


http://www.uniroma3.6net.garr.it/mrtg/graphs/gsr.html

Other remarks:

IPv6 must be explicitly enabled in the configuration file (MRTG) or via command-line options (cfgmaker).

IPv6 support requires the Socket6 (v0.11 or later) and IO::Socket::INET6 (v1.26 or later) Perl modules, which can be downloaded from CPAN (http://www.cpan.org/).

Tests description

MRTG’s IPv6 support was tested with Net-SNMP agents under Linux and with a Cisco GSR router running an EFT image with support for SNMP over IPv6. Regression tests were also run, on a Cisco 2600 and a Cisco 7500, to investigate whether IPv6 support causes problems with the monitoring of dual stack routers that do not support SNMP over IPv6 and must be queried over IPv4.

Test environment description (schemes, topology, configuration)

46

http://people.ee.ethz.ch/~oetiker/webtools/mrtg/

http://www.uniroma3.6net.garr.it/mrtg/graphs/gsr.html

http://www.cpan.org/


The tests were run on various Linux distributions including Red Hat 7.2 and 8.0, Debian unstable, and Mandrake 9.1. All hosts were dual stack. The IPv6 libraries used were Socket6.pm v0.11 and INET6.pm v1.26. No special network or host configuration was performed. Test description, results (including screenshots, frame captures), observed difficulties Testing consisted in using cfgmaker to generate MRTG configuration files for all routers, using both IPv6 and IPv4, and using MRTG to monitor traffic and generate traffic graphs using the generated configuration files. In all cases, IPv6 was enabled by passing the --enable-ipv6 option to cfgmaker. Both numeric IPv6 addresses and hostnames were tried.

Test summary, remarks All tests were successful. In the case of dual stack routers that did not support IPv6, cfgmaker correctly switched to IPv4 when SNMP queries over IPv6 failed. However, as MRTG does not currently support IPv6 MIBs, IPv6 interface addresses were not displayed in the generated web pages. Also, more testing is required to determine if IPv6 support works on other operating systems such as FreeBSD.

47


Name of the tool:




available:

Multicast Beacon PSNC version

All OS platform with Java (v.1.4 and higher)

GPL

NALNR (original version), http://dast.nlanr.net/Projects/Beacon/

[email protected]

PSNC (PSNC version)

http://noc.man.poznan.pl/noc/strony/aplikacje.html

[email protected]

Yes


Roman Łapacz, PSNC

[email protected] Yes

IPv6 status:

IPv6 enabled


The Multicast Beacon is a measurement tool to monitor multicast traffic.Its architecture is client-server. Clients exchange test packets between themselves using multicast technology. This way they compute traffic parameters (delay, loss, jitter, order, duplicate) and send gathered results to the Beacon server. The server offers web presentation to the end user.


The application is ready to use but still is being developed


DE

VE

LO

PME

NT


http://beaconserver.m6bone.pl/

Other remarks:

Tests description


48


http://noc.man.poznan.pl/noc/strony/aplikacje.html


Test environment description (schemes, topology, configuration) Multicast: present in the network OS platform: Linux Java version: 1.4 (IPv6 support) Nr of clients: more then one in different locations Executive script of Beacon client: -- multicast_group=FF0e::8320:1 multicast_port=56465 beacon_server=3ffe:8320:5:101:210:4bff:fe91:9120 server_port=19996 time_to_live=127 user_name=PSNC ementation on your platform /usr/java/j2sdk1.4.0/bin/java -client Beacon \ -group $multicast_group \ -port $multicast_port \ -server $beacon_server \ -serverport $server_port \ -ttl $time_to_live \ -name $user_name Executive script of Beacon server: beacon_port=19996 # beacons report to this port viewer_port=19998 # viewer send query to this port web_view_port=80 # web browser send query to this port default_session=FF0e::8320:1:56465 # default session for web view (shortcut) history_interval=10 # interval of statistics archiving /usr/java/j2sdk1.4.0/bin/java -server -cp "/usr/java/j2sdk1.4.0/jre/lib/rt.jar:./class:./lib/freemarker152.jar:./l ib/mail.jar:./lib/activation.jar" BeaconServer \ -beaconport $beacon_port \ -viewerport $viewer_port \ -webviewport $web_view_port \ -defaultsess $default_session \ -history $history_interval


49


Test summary, remarks The application is stable and runs in both network infrastructures: IPv6 and Ipv4.

50




License type:



Nagios

Nagios version 1.1and Nagios

plugins 1.3

Linux, *BSD and other UNIX operating system

GPLv2 www.nagios.org Yes


János Mohácsi (HUNGARNET) Yes

IPv6 status:

The IPv6 support depends on several factors.

- The information reporting is relying on an external webserver. If the webserver supports IPv6 the visualization part supports IPv6.

- The ping plugin can be interfaced for a while to use fping that supports IPv6 therefore IPv6 support for pinging is available in the current version.

- Other check plugins has recently extended to support IPv6. Next version of plugins (v1.4) will support IPv6 service monitoring.


Nagios is a host and service monitor designed to inform network operators about the network problems. The monitoring daemon runs intermittent checks on hosts and services you specify using external "plugins" which return status information to Nagios. When problems are encountered, the daemon can send notifications out to administrative contacts in a variety of different ways (email, instant message, SMS, etc.). Current status information, historical logs, and reports can all be accessed via a web browser.


For Nagios probably version 2.0 and

For Nagios plugins version 1.4

Planned extensions, new features ,with dates: DE

VE

LO

PME

NT

Internationalization, grouping services, more rational configuration files


NIIF/HUNGARNET: http://6net.iif.hu/nagios/

Other remarks:

Tests description

The NAGIOS software is used at NIIF/HUNGARNET for monitoring networking services. We set up a separate NAGIOS monitoring station for monitor IPv6 network service of HUNGARNET and 6NET also. We wanted to test the feasibility of using NAGIOS for monitoring IPv6 services. Find out the missing functionality and develop the missing components.

51

http://6net.iif.hu/nagios/


Test environment description (schemes, topology, configuration) We tested NAGIOS on a FreeBSD 4.8-STABLE but also some features also tested on Solaris 8.The NAGIOS was installed from the port collection: Nagios version 1.0 later 1.1 and Nagios plugins 1.3beta2 and later 1.3 No special network and host configuration performed. Test description, results (including screenshots, frame captures), observed difficulties Initially the NAGIOS configured to monitor IPv6 reachability of 6NET core routers and 6NET ping hosts and NIIF/HUNGARNET IPv6 enabled routers. To be able to test IPv6 reachability from NAGIOS we had to use check_fping plugin with IPv6 enabled fping since check_ping did not support IPv6. After our report and help IPv6 support for check_ping appeared in the CVS tree. Then we tested IPv6 support of service checking plugins. We found to be unavailable and therefore we developed code for it. We reported to nagios plugins developers, and they implemented a more rational interface of IPv6 service checking, that is available in the CVS tree. We tested the CVS version and we found a name resolution handling error. After our bug report it has been corrected. In the next upcoming release (v1.4) Nagios plugins will have fill IPv6 service checking support.

Test summary, remarks The test of Nagios CVS version was successful. The tool will be IPv6 enabled soon. However more service testing and other platform testing would be necessary.

52


Name of the tool:



License type:



Netflow/IPFIX Netflow v9

List of supported platforms

open specification

http://www.cisco.com/go/netflow/, http://ipfix.doit.wisc.edu/ No


Simon Leinen (SWITCH), Janos Mohacsi (HUNGARNET) Yes (SWITCH, SURFNet, HUNGARnet, DANTE)

IPv6 status:

IPv6 Support: None in versions prior to v9. An initial router implementation of IPv6 support for Netflow v9 is now available as an IOS EFT (Early Field Test, NetFlowIos). A few collectors have added IPv6 support, mostly in experimental status (NetFlowUtc).


Netflow is a flow-based traffic accounting protocol defined by Cisco Systems. It is widely used to support various applications such as usage-based charging, traffic analysis, or capacity planning. The latest version, Netflow v9, will be used as a basis for the IPFIX (IP Flow Information eXport) protocol that is currently being standardized in the IETF



DE

VE

LO

PME

NT

Required extensions to be operational within a native IPv6 environment

• Transport of Netflow accounting data over IPv6

Proposed extensions

• Support for the emerging IPFIX protocol

• Metering/export monitoring through a Netflow MIB




Other remarks:

Application domains in the 6Net management framework Network Element usage data collection, Network Element trend analysis, Network Element capacity planning, Traffic management, network capacity analysis, network data aggregation and trending, network usage correlation, network usage data storage, traffic pattern analysis, network security breach detection, traffic reporting, customer profiling, usage pattern analysis, service intrusion detection, feature pricing, usage charge, correlation, validation and aggregation.

53

http://www.cisco.com/go/netflow/

http://ipfix.doit.wisc.edu/

http://6nettools.dante.net/cgi-bin/moin/moin.cgi/NetFlowIos

http://6nettools.dante.net/cgi-bin/moin/moin.cgi/NetFlowUtc



Name of the tool:



License type:



Net-snmp Version 5.0.3, 5.0.8

All major unices + windows

Open source

www.net-snmp.sourceforge.net for the net-snmp offical version

w6.loria.fr for the extended version

Yes


For version 5.0.3: Isabelle ASTIC, Olivier FESTOR

for version 5.0.8: Athanasios Douitsis [email protected] No

IPv6 status:

The net-snmp project offers an IPv6 support to the SNMP protocol and for the first MIBs defined by IETF in 1996, which was able to manage IPv6 networks only.


The net-snmp project was originally based on the Carnegie Mellon University and Univerisity of California at Davis SNMP implementations .It has various tools relating to the Simple Network Management Protocol including:

• An extensible agent

• An SNMP library

• tools to request or set information from SNMP agents

• tools to generate and handle SNMP traps

• a version of the unix 'netstat' command using SNMP

The tools provide to query and act on remote SNMP agents are :

• snmpget command, which can be used to retrieve data from a remote host given its host name, authentication information and an OID.

• snmpgetnext command, which is similar in usage to the snmpget command, is used to retrieve the next OID in the MIB tree of data. Instead of returning the data requested, it returns the next OID in the tree and its value:

• The snmpwalk command which essentially performs a whole series of getnext commands automatically, and stops when it returns results that are no longer inside the range of the OID that was originally specified. This command can be used to get all of the information stored on a machine in the system MIB group.


Package is constantly under developement by many people around the world on a voluntary basis. Next version number will probably be 5.0.9 or something similar.

DE

VE

LO

PM

EN

T

Possible extensions:

54

http://www.net-snmp.sourceforge.net/


Some extensions can be done, like notification support or the udpListenerTable or the TcpListenerTable, now that FreeBSD kernel manages UDP v6 and TCP v6. It is thus possible to extend our current implementation.


Other remarks:


Our extension of the net-snmp package implements the new MIB II which is able to manage both IPv4 and IPv6 networks. At the time of the development, the implementation concerned draft-ietf-ipngwg-rfc2011-update-00.txt,

draft-ietf-ipngwg-rfc2012-update-01.txt, draft-ietf-ipngwg-rfc2013-update-01.txt, draft-ietf-ipngwg-rfc2096-update-00.txt.

Within the MIB directories, four files have been added (one per draft) and the INET-ADDRESS-MIB.txt file

has been updated with the RFC3291.

Those drafts are only implemented in read access. No set function is currently defined, neither notifications. Some tables are not yet complete, as udpListenerTable, because the system did not manage the information we needed, at the time of development.

This implementation has been done on FreeBSD 4.5-RELEASE #22

Test description #1 performed by Isabelle ASTIC, Olivier FESTOR

During the development, tests were made to verify the consistency of the data returned.

Test environment description (schemes, topology, configuration) The tests were made on the IPv6 testbed platform of the LORIA. All systems were defined with FreeBSD 4.5 RELEASE #22-. Two nodes were used, one to support the net-snmp manager, the other the net-snmp agent. Test description, results (including screenshots, frame captures), observed difficulties

It is not of a great interest to put here the entire MIB II that we implemented. Just as an example, here is an extract of the output for ipAddressTable : IP-MIB::ipAddressAddrType.ipv4.32.127.0.0.1 = INTEGER: ipv4(1) IP-MIB::ipAddressAddrType.ipv4.32.152.81.48.2 = INTEGER: ipv4(1) IP-MIB::ipAddressAddrType.ipv6.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = INTEGER: ipv6(2) IP-MIB::ipAddressAddrType.ipv6.128.32.1.6.96.3.1.0.50.2.1.2.255.254.227.96.138 = INTEGER: ipv6(2) IP-MIB::ipAddressAddrType.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = INTEGER: ipv6(2) IP-MIB::ipAddressAddrType.ipv6.128.254.128.0.0.0.0.0.0.2.1.2.255.254.227.96.138 = INTEGER: ipv6(2) IP-MIB::ipAddressAddr.ipv4.32.127.0.0.1 = STRING: "127.0.0.1" IP-MIB::ipAddressAddr.ipv4.32.152.81.48.2 = STRING: "152.81.48.2"

55


IP-MIB::ipAddressAddr.ipv6.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = STRING: "::1" IP-MIB::ipAddressAddr.ipv6.128.32.1.6.96.3.1.0.50.2.1.2.255.254.227.96.138 = STRING: "2001:660:301:32:201:2ff:fee3:608a" IP-MIB::ipAddressAddr.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = STRING: "fe80::1" IP-MIB::ipAddressAddr.ipv6.128.254.128.0.0.0.0.0.0.2.1.2.255.254.227.96.138 = STRING: "fe80::201:2ff:fee3:608a" IP-MIB::ipAddressIfIndex.ipv4.32.127.0.0.1 = INTEGER: 3 IP-MIB::ipAddressIfIndex.ipv4.32.152.81.48.2 = INTEGER: 1 IP-MIB::ipAddressIfIndex.ipv6.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = INTEGER: 3 IP-MIB::ipAddressIfIndex.ipv6.128.32.1.6.96.3.1.0.50.2.1.2.255.254.227.96.138 = INTEGER: 1 IP-MIB::ipAddressIfIndex.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = INTEGER: 3 IP-MIB::ipAddressIfIndex.ipv6.128.254.128.0.0.0.0.0.0.2.1.2.255.254.227.96.138 = INTEGER: 1 IP-MIB::ipAddressType.ipv4.32.127.0.0.1 = INTEGER: unicast(1) IP-MIB::ipAddressType.ipv4.32.152.81.48.2 = INTEGER: unicast(1) IP-MIB::ipAddressType.ipv6.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = INTEGER: unicast(1) IP-MIB::ipAddressType.ipv6.128.32.1.6.96.3.1.0.50.2.1.2.255.254.227.96.138 = INTEGER: unicast(1) IP-MIB::ipAddressType.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = INTEGER: unicast(1) IP-MIB::ipAddressType.ipv6.128.254.128.0.0.0.0.0.0.2.1.2.255.254.227.96.138 = INTEGER: unicast(1) IP-MIB::ipAddressPrefix.ipv4.32.127.0.0.1 = OID: IP-MIB::ipAddressPrefixPrefix.3.ipv4.32.127.0.0.0.8 IP-MIB::ipAddressPrefix.ipv4.32.152.81.48.2 = OID: IP-MIB::ipAddressPrefixPrefix.1.ipv4.32.152.81.48.0.24 IP-MIB::ipAddressPrefix.ipv6.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = OID: IP-MIB::ipAddressPrefixPrefix.3.ipv6.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.1.128 MIB::ipAddressPrefixPrefix.1.ipv6.128.32.1.6.96.3.1.0.50.0.0.0.0.0.0.0.0.64 IP-MIB::ipAddressPrefix.ipv6.128.32.1.6.96.3.1.0.50.2.1.2.255.254.227.96.138 = OID: IP-MIB::ipAddressPrefixPrefix.1.ipv6.128.32.1.6.96.3.1.0.50.0.0.0.0.0.0.0.0.64 MIB::ipAddressPrefixPrefix.2.ipv6.128.32.1.6.96.3.1.0.51.0.0.0.0.0.0.0.0.64 MIB::ipAddressPrefixPrefix.2.ipv6.128.32.1.6.96.3.1.0.51.0.0.0.0.0.0.0.0.64 IP-MIB::ipAddressPrefix.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.1 = OID: IP-MIB::ipAddressPrefixPrefix.3.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.64 MIB::ipAddressPrefixPrefix.2.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.64 IP-MIB::ipAddressPrefix.ipv6.128.254.128.0.0.0.0.0.0.2.1.2.255.254.227.96.138 = OID: IP-MIB::ipAddressPrefixPrefix.1.ipv6.128.254.128.0.0.0.0.0.0.0.0.0.0.0.0.0.0.64 The most difficult part was to obtain the information from the kernel. Sometimes, like for UDP or TCP table, the information was not managed by the kernel. Sometimes, it was partially managed, like the data for the INET ICMP Table, where we could only find the information for the system-wide and not interface per interface

Tests description #2 performed by Athanasios Douitsis


Test environment description (schemes, topology, configuration) The test machine was an i386 machine running Linux 2.4.x. The package was downloaded, compiled and tested in that machine. The IPV6 stack was enabled and was using stateless autoconfiguration. Other than that, no special configuration or customization was made. The package configuration script includes a handy –enable-ipv6 option which allows the administrator to enable all available ipv6-aware mechanisms with one argument. There are a lot of available mib modules that can be compiled if the administrator desires it, but very few of them were of actual value during this test. The package compiled and installed perfectly without any trouble whatsoever. The snmpd.conf file supports ipv6 equivalents of all usefull operands. For example, the operand rocommunity has its rocommunity6 counterpart for ipv6 transports. The ability to create elaborate access lists and VACM configurations is present in the ipv6 domain also.


56


The primary objective of the tests was to ascertain whether the agent has or has not the ability to use ipv6 as the transport for the snmp communication with other management entities. Since the Net-SNMP package offers a full range of snmpget, snmpwalk etc commands, it was also imperative that those tools' ability to use ipv6 be tested too. After the software was installed, a sample configuration was created, allowing only ipv6 transport capabilities and the agent was launched. We tested the agent by doing several snmp queries on various sub trees of the OID tree. All kinds of snmpwalks , snmpgets etc were successful so it was clear that the ipv6 transport capabilities of the agent are quite complete.

Test summary, remarks Overall impression is that the Net-SNMP package has excellent IPV6 support and is clearly ready for widespread use. The only bad thing is that there are only a few MIBs that include information related to IPV6 and even those seem to be probably incomplete. For example, the ipv6-mib lacked several objects which according the ASN.1 file should be there.

57




License type:



Network Weathermap N/A All major Unix systems with Perl support

To be released under GPL



Athanassios Liakopoulos ([email protected]) Yes (http://netmon.grnet.gr/6net.html)

IPv6 status: The Network Weathermap tool is a perl script that uses direct (IPv6/IPv4) SNMP calls to gather data from appropriate router interface MIB objects.


The “Network Weatermap” tool is an SNMP-based network traffic condition grapher. Network connections are depicted as colour arrows usually over geographical maps. The colour of the arrow changes according to the current load of the network connection. Placing the pointer over the arrows, the MRTG graphed line utilisation is presented while double-clicking the on the arrow, the corresponding MRTG traffic analysis page is presented. The “Network Weathermap” front page is updated every five-minute interval


There is no new functionality added to the tool. However, the code is reconstructed and documented in order to facilitate the deployment of the software. The next version will be 1.0 and its expected to be released in end of July 2003Planned extensions, new features ,with dates:

DEVELOPMENT

Not available at this time


6NET network traffic load is depicted in http://netmon.grnet.gr/6net.html. Additional, GRNET network traffic load may be found in http://netmon.grnet.gr/

Other remarks:

The Network Weathermap code is currently released but it is available if requested.

Tests description

Test environment description (schemes, topology, configuration) Network Weathermap is running as a service for the 6NET/GRNET network since the beginning of 2003. Earlier (IPv4-only) versions of the tool were tested in the past in order to validate the perl script code. After adding IPv6 support to the tool, limited tests were performed as IPv6 calls with available SNMP tools were validated from other research groups. The tool is tested in the past in many different Unix environments.


58

http://netmon.grnet.gr/6net.html


Follows a screen shot of theNnetwork Weathermap tool showing the 6NET network.

Test summary, remarks The Network Weatermap may be used for building production services in IPv6 only or dual stack networks.

59


Name of the tool:




available:

PCHAR 1.4

06.13.2001

FreeBSD, NetBSD, Solaris, Linux, OSF/1

GPL http://www.employees.org/~bmah/Software/pchar

Bruce A. Mah Yes


PSNC: Lucas Dolata, Wiktor Procyk Yes

IPv6 status:

To enable IPv6 support, give the --with-ipv6 option to configure. If there is a directory for IPv6-specific libraries, it can be specified via an argument to the --with-ipv6 option, for example:

% ./configure --with-ipv6=/usr/local/v6

% make


Pchar is a tool to characterize the bandwidth, latency, and loss of links along an end-to-end path through the Internet. It is based on the algorithms of the pathchar utility written by Van Jacobson, formerly of Lawrence Berkeley Laboratories.

Pchar measures the characteristics of the network path between two Internet hosts, on IPv4 or IPv6 network.

The program measures network throughput and round trip time by sending varying sized UDP packets into the network and waiting for ICMP messages in response. It modulates the IPv4 time to live (TTL) field or the IPv6 hop limit field to get measurements at different distances along a path.

Pchar for each hop in the trip shows the following details:

• the number of partial lost datagrams and percentage of probe packets that were lost during the probes for that hop

• the estimated round trip time from the probing host through the current hop

• estimates of the round trip time and bandwidth for the current hop

• estimate of the average queuing along the path, up to and including the current hop

After the last hop (usually the target host), pchar prints statistics in the entire path, including the path length and path pipe (the latter is an estimate of the delay bandwidth product of the path).

In the other/second mode of operation called trout (short for “tiny traceroute”). Pchar sends packets of random sizes (one packet per hop diameter) along the path to a destination. This mode is extremely fast but no attempt at estimating link properties is made.


Software is not under development

DE

VE

LO

PME

NT


60

http://www.employees.org/~bmah/Software/pchar

http://www.employees.org/~bmah/


N/a


There have been created special site where you can try use Pchar in your research

http://www.ipv6.man.poznan.pl/cgi-bin/pchar.cgi

Other remarks:

N/a

Tests description


Pchar do not need special network configuration, host configuration and network topology. It can be used on every machine. In test there have been used Redhat 8.0 Linux and Suse 8.0 Linux. To compile the downloaded version of pchar in some Linux distributions (i.e. RedHat 8.0) it is necessary to make a few changes in the source codes. The compliance problem occur because there are differences in the udphdr structure, between various unix systems. Test description, results (including screenshots, frame captures), observed difficulties Bellow there are shown some tests which were made using PCHAR. There were tested paths between host in Poland boletus.man.poznan.pl and some host from 6net core network:

pchar to lo0.se6.se.6net.org (2001:798:25::1) using UDP/IPv6 Using raw socket input Packet size increments from 52 to 1000 by 32 31 test(s) per repetition 3 repetition(s) per hop 0: 2001:808::2 (2001:808::2) Partial loss: 0 / 90 (0%) Partial char: rtt = 1.105496 ms, (b = 0.000990 ms/B), r2 = 0.722094 stddev rtt = 0.123493, stddev b = 0.000116 Partial queueing: avg = 0.000639 ms (645 bytes) Hop char: rtt = 1.105496 ms, bw = 8080.983999 Kbps Hop queueing: avg = 0.000639 ms (645 bytes) 1: 2001:808::7 (2001:808::7) Partial loss: 0 / 90 (0%) Partial char: rtt = 24.949370 ms, (b = 0.001405 ms/B), r2 = 0.790427 stddev rtt = 0.145433, stddev b = 0.000137 Partial queueing: avg = 0.000638 ms (645 bytes) Hop char: rtt = 23.843874 ms, bw = 19293.139293 Kbps Hop queueing: avg = -0.000001 ms (0 bytes) 2: 2001:798:25:201::1 (2001:798:25:201::1) Path length: 2 hops Path char: rtt = 24.949370 ms r2 = 0.790427 Path bottleneck: 8080.983999 Kbps Path pipe: 25201 bytes Path queueing: average = 0.000638 ms (645 bytes) Start time: Thu Jun 5 09:51:53 2003 End time: Thu Jun 5 09:52:40 2003 pchar to lo0.de6.de.6net.org (2001:798:14::1) using UDP/IPv6 Using raw socket input Packet size increments from 52 to 1000 by 32 31 test(s) per repetition 3 repetition(s) per hop 0: 2001:808::2 (2001:808::2) Partial loss: 0 / 90 (0%) Partial char: rtt = 1.284268 ms, (b = 0.000781 ms/B), r2 = 0.590162 stddev rtt = 0.130818, stddev b = 0.000123

61

http://www.ipv6.man.poznan.pl/cgi-bin/pchar.cgi


Partial queueing: avg = 0.000486 ms (622 bytes) Hop char: rtt = 1.284268 ms, bw = 10247.203818 Kbps Hop queueing: avg = 0.000486 ms (622 bytes) 1: 2001:808::7 (2001:808::7) Partial loss: 0 / 90 (0%) Partial char: rtt = 25.325033 ms, (b = 0.001032 ms/B), r2 = 0.634418 stddev rtt = 0.157546, stddev b = 0.000148 Partial queueing: avg = 0.001017 ms (2735 bytes) Hop char: rtt = 24.040765 ms, bw = 31816.409760 Kbps Hop queueing: avg = 0.000531 ms (2113 bytes) 2: 2001:798:25:201::1 (2001:798:25:201::1) Partial loss: 1 / 90 (1%) Partial char: rtt = 38.986758 ms, (b = 0.001277 ms/B), r2 = 0.645145 stddev rtt = 0.190410, stddev b = 0.000179 Partial queueing: avg = 0.000902 ms (2735 bytes) Hop char: rtt = 13.661725 ms, bw = 32695.553346 Kbps Hop queueing: avg = -0.000115 ms (0 bytes) 3: 2001:798:0:8::2 (se.de6.de.6net.org) Path length: 3 hops Path char: rtt = 38.986758 ms r2 = 0.645145 Path bottleneck: 10247.203818 Kbps Path pipe: 49938 bytes Path queueing: average = 0.000902 ms (2735 bytes) Start time: Thu Jun 5 11:50:36 2003 End time: Thu Jun 5 11:51:52 2003 pchar to lo0.uk6.uk.6net.org (2001:798:28::1) using UDP/IPv6 Using raw socket input Packet size increments from 52 to 1000 by 32 31 test(s) per repetition 3 repetition(s) per hop 0: 2001:808::2 (2001:808::2) Partial loss: 0 / 90 (0%) Partial char: rtt = 1.103903 ms, (b = 0.000999 ms/B), r2 = 0.774260 stddev rtt = 0.108488, stddev b = 0.000102 Partial queueing: avg = 0.000678 ms (678 bytes) Hop char: rtt = 1.103903 ms, bw = 8006.289702 Kbps Hop queueing: avg = 0.000678 ms (678 bytes) 1: 2001:808::7 (2001:808::7) Partial loss: 0 / 90 (0%) Partial char: rtt = 24.966060 ms, (b = 0.001056 ms/B), r2 = 0.601759 stddev rtt = 0.172701, stddev b = 0.000162 Partial queueing: avg = 0.000559 ms (678 bytes) Hop char: rtt = 23.862158 ms, bw = 141452.980947 Kbps Hop queueing: avg = -0.000118 ms (0 bytes) 2: 2001:798:25:201::1 (2001:798:25:201::1) Partial loss: 0 / 90 (0%) Partial char: rtt = 53.804452 ms, (b = 0.001450 ms/B), r2 = 0.850281 stddev rtt = 0.122365, stddev b = 0.000115 Partial queueing: avg = 0.000747 ms (1153 bytes) Hop char: rtt = 28.838392 ms, bw = 20280.756792 Kbps Hop queueing: avg = 0.000188 ms (475 bytes) 3: 2001:798:0:7::1 (se.uk6.uk.6net.org) Path length: 3 hops Path char: rtt = 53.804452 ms r2 = 0.850281 Path bottleneck: 8006.289702 Kbps Path pipe: 53846 bytes Path queueing: average = 0.000747 ms (1153 bytes) Start time: Thu Jun 5 14:29:38 2003 End time: Thu Jun 5 14:30:52 2003


Tests which have been made shown the main disadvantage of pchar tool. The accuracy of determining measures depends on network infrastructure. Pchar use algorithms of pathchar to characterize the bandwidth, latency, and loss of links along an end-to-end path through the Internet. In this algorithm the differences between measures and real value of parameters rise, if links which have less bandwidth precede links which have more bandwidth.

62


Name of the tool:



License type:



Polyphemus 1.0 Cross-platform (Java) freeware www.dia.uniroma3.it/~polyph No


Roma Tre Computer Networks research group No IPv6 status:

IPv6 is not yet supported.


Polyphemus works only on networks that use the OSPF protocol and exploits the protocol information for learning the network topology. Is is based on a two levels architecture with a client and a drawing-server.

Polyphemus requires to access private routing information using SNMP. It is conceived as a local application that has to be installed and used by single users. The drawing-server of Polyphemus encapsulate a graph drawing engine that computes visual maps of the network exploiting sophisticated algorithms.


1.1


DE

VE

LO

PME

NT

(unknown date) – IPv6 MIB and OSPFv3 support


http://www.dia.uniroma3.it/~polyph/PolyphemusClip.zip

Other remarks:

IPv6 support is planned in the near future. The first stage will probably be support for the IPv6 MIBs and OSPFv3, and support for SNMP over IPv6 will come later.

Tests description

No tests have been carried out as Polyphemus does not yet support IPv6.

63

http://www.dia.uniroma3.it/~polyph


Name of thetool:


Available platforms: License type:

Tool location andinformation links,authors:


RANCID 2.2.2

Any UNIX like operating system where CVS available. Rancid currently supports Bay routers, Cisco routers, Juniper routers, Catalyst switches, Foundry switches, Redback NASs, ADC EZT3 muxes, MRTd (and thus likely IRRd), Alteon switches, and HP procurve switches

Freely available copyrighted

http://www.shrubbery.net/rancid/ Yes


János Mohácsi (NIIF/HUNGARNET), Ioannis Kappas (DANTE), Rob Evans (UKERNA), Christian Schild,Westfälische Wilhelms-Universität Münster (JOIN)

Yes

IPv6 status:

Rancid itself is not aware of IPv6, but it uses external programs It is relying on the telnet, ssh or rlogin tools of host operating system to login and obtain configuration information.. If these remote access programs are IPv6-capable one can use IPv6 addresses and transport


RANCID - Really Awesome New Cisco confIg Differ (http://www.shrubbery.net/rancid/) is a tool written in perland it is a configuration management tool for Cisco routers and Catalyst switches, as well as equipment from Alteon, Bay Networks, Extreme, Force 10 Networks, Foundry, HP, Juniper and Redback. It works by periodically connecting to devices by means of telnet, SSH, or rlogin and recording the configuration. Any differences are flagged using diff and emailed to network operator and saved in CVS. With CVS, changes over time in these configurations can be tracked. There are various frontends to watch these changes conveniently (e.g."cvsweb" or"viewcvs").


Probably 2.2.3


DEVELOPMENT

Extension to support other router platform (done for HITACHI at HUNGARNET)

make it more modular (ongoing work at HUNGARNET)


6NET core: http://tools.6net.org/cgi-bin/cvsweb.cgi/ NIIF/HUNGARNET: http://6net.iif.hu/routerconfig/

Other remarks:

64

http://www.shrubbery.net/rancid/

http://6net.iif.hu/routerconfig/

http://6net.iif.hu/routerconfig/


Tests description

The managers and operators of 6NET can look to the 6NET core RANCID webpages to find out the current configuration and latest configuration changes of core 6NET routers. They can use it for operational management to configuration management of the 6NET core routers. At NIIF/HUNGARNET we are also using for similar purpose to provide configuration information to network managers of the IPv6 enabled regional centers.

Test environment description (schemes, topology, configuration) At NIIF/HUNGARNET we installed RANCID 2.2.2 on a FreeBSD 4.8-STABLE host from the port collection. No special software and network configuration required except adding name and type of the router we wanted to monitor router into the configuration file of RANCID. We also added a special account for monitoring purpose on the FreeBSD monitoring host.

Test description, results (including screenshots, frame captures), observed difficulties We tested RANCID in an IPv6 only and IPv6/IPv4 dualstack environment. We tested numerical address and hostname also.

Test summary, remarks All tests were successful. We have to test the ssh/rlogin interface also since currently we used telnet interface. More testing is required to determine if IPv6 support works on other operating systems such as Linux. Some testing also done on Solaris in the 6NET core sites.

65


Name of the tool:



License type:



RIPE NCC Test Traffic Server with IPv6 support

February 2003

Dedicated FreeBSD based system supplied by RIPE NCC

System can be bought and maintained by RIPE NCC. Cost currently under review

http://www.ripe.net/ttm//ttm-ipv6.html No


Tim Chown, University of Southampton Yes

IPv6 status:

IPv6 functionality was added to the existing RIPE NCC TT server after discussion between 6NET and the RIPE NCC which started in Q1 2002. It was decided that RIPE NCC rather than 6NET would develop the new IPv6 functionality so that expertise could be gathered and maintained in RIPE NCC.


The TT server allows statistics to be gathered between any pair of deployed TT servers. The statistics include packet delay and loss, as well as a historical view of observed traceroutes. The system is available as a “black box” shipped from RIPE-NCC. It requires a roof-mounted GPS to be deployed for time synchronisation. Statistics are gathered at the RIPE NCC site and presented for views there by RIPE-NCC TT server owners (once you own a box, you can view any details). There is a purchase fee and maintenance fee – these fees are currently under review and likely to be lowered (purchase is around 3,000 Euros, maintenance is likely to fall to 1,000 Euros p.a.).


RIPE NCC is still working on new IPv6 functionality, e.g. AS path tracing alongside IP-based traceroute views


DE

VE

LO

PME

NT

To be determined by RIPE NCC.


http://www.ripe.net/ttm/Plots/IPv6/

(requires user name and password)

Other remarks:

The TT server is a well-established tool for IPv4 (around 70 deployed monitoring systems), which is now shipping with IPv6 support by default as a result of 6NET’s initiative to promote adoption of IPv6 by RIPE NCC.

66


Tests description

Here we illustrate the usage of the TT server by showing plot results.

The function to view delay and loss plots for the previous day, week, month and 6 month period is working.

The traceroute view function is working (but the AS view is not yet implemented for IPv6).

Test environment description (schemes, topology, configuration) The TT server data views are available via a web server at the RIPE NCC web site, protected by username and password. Local statistics, and details such as GPS satellite reception quality, can be viewed via the local TT server web server. This does not yet support IPv6 access, but this should be available soon. Test description, results (including screenshots, frame captures), observed difficulties The following three figures show:

a. The overview “dashboard” view of performance for the last day, week, month and 6 months. From this the interesting plots or trends can be seen, and the more detailed chart selected.

b. Here we are focused on the month view from the above page. The top left chart shows

scatter of delays, with the red line being an indication of the hop count between the hosts. The delays distribution can be seen on the right.

c. In this chart we see part of the historic traceroute view. This is very useful for diagnosing

routing path problems. All three plots are from the live system between University of Southampton, UK and NTUA, Greece.

67


68


Test summary, remarks The plots are proving very useful for IPv6 monitoring, especially the historical traceroute view (while we are at the stage of routing experiments and occasional instability).

69


70


Name of the tool:



License type:



SNMP_Session.pm 0.97 artistic http://www.switch.ch/misc/leinen/snmp/perl/ Yes


Simon Leinen, SWITCH No

IPv6 status:

IPv6 Support: YES


An SNMP library written entirely in Perl. There are two layers of programming interface: SNMP_Session.pm/BER.pm provide a low-layer interface to encoding and decoding of SNMP requests and data structures. SNMP_util.pm adds basic MIB parsing support and convenient interfaces to some basic SNMP operations.



DE

VE

LO

PME

NT

Required extensions to be operational within a native IPv6 environment In order to communicate with managed devices over IPv6, support for INET6 address family sockets is needed. The package initially used the IO::Socket::INET class. A usable version of a IO::Socket::INET6 module has very recently been contributed to CPAN by Rafael Martinez Torres (Novagnet). IPv6 support for SNMP_Session.pm was independently developed by Rafael on one hand and by Lorenzo Colitti (Universita Roma III) [MrtgV6Tool] on the other. The Colitti implementation seemed to be put a higher concern on backward compatibility, so it has been selected for inclusion in MRTG and SNMP_Session.


Other remarks:

Application domains in the 6Net management framework This is a library and cannot be used directly. However, it has been used to support applications in network capacity planning (MRTG, Cricket) network and service monitoring (devpatrol, Big Sister), inventory management and other areas.

71

http://www.switch.ch/misc/leinen/snmp/perl/

http://6nettools.dante.net/cgi-bin/moin/moin.cgi/MrtgV6Tool




License type:


Is sourceavailable:

Westhawk's Java SNMP stack

4.13

20/11/2002 JDK 1.3.0

The stack is free and commercial support isavailable

http://www.westhawk.co.uk/resources/snmp/index.html ,

Tim Panton & Birgit Arkesteijn,

[email protected]

Yes


Ioannis Kappas (DANTE), Robert Szuman (PSNC), Bartosz Belter (PSNC) No

IPv6 status:

Officially, in the current version of the SNMP stack the IPv6 protocol is not supported. In spite of this, some simple tests performed in PSNC under a native IPv6 environment proved that the tested stack supports the IPv6 transport. It worked correctly under JDK 1.4.1.


It is a lightweight implementation of Java SNMP stack, with Java application and servlet examples. The stack provides manager functionality for SNMPv1, SNMPv2c and SNMPv3 (both authentication and privacy). As authentication protocols the stack offers MD5 and SHA1. It is capable of sending and receiving traps, but has no other agent functionality. This is the “lightweight” type package, so the target applications can be small applets, which can be easily popped in any frame (e.g. Netscape). This stack has no MIB browsing capabilities, so the application has to know the OID (Object Identifier) of MIB parameters it wants to monitor or it can use any other packages that do MIB browsing and general network management. The source code and documentation is included in the distribution.


The project is still open and under development.


DE

VE

LO

PME

NT

The use of Java 1.4 features to improve support polling of large numbers of agents.

Support for agent functionality (if enough manpower).


MUVI (Multicast Visualisation Tool) http://muvi.man.poznan.pl

Other remarks:

A much more detailed description about the architecture of this SNMP stack and the history of its design and development can be found in „The Simple Times” magazine (the December 2001 issue) and is accessible via Web at http://www.simple-times.org/pub/simple-times/issues/9-1.html#tools

72

http://www.westhawk.co.uk/resources/snmp/index.html




http://muvi.man.poznan.pl/

http://www.simple-times.org/pub/simple-times/issues/9-1.html


Tests description


The main objective of this test is to provide information about IPv6 transport support in Westhawk's Java SNMP stack.

baobab.man.poznan.pl2001:808:0:6:207:e9ff:fee3:55d5

hollyhock.man.poznan.pl2001:808:0:6:202:3fff:feaf:9d52

SNMPGET

SNMP REPLY

Westhawk's JavaSNMP stack

NET-SNMPagent

Fig. 6 Test topology

Configuration: - hollyhock.man.poznan.pl

address: 2001:808:0:6:202:3fff:feaf:9d52 platform: Linux application: MUVI MIB Browser with Westhawk’s Java SNMP stack

- baobab.man.poznan.pl address: 2001:808:0:6:207:e9ff:fee3:55d5 platform: Linux snmp agent: NET-SNMP

Test description, results (including screenshots, frame captures), observed difficulties hollyhock.man.poznan.pl station performed the SNMPGET operation using the MUVI MIB Browser.

The NET-SNMP agent located at baobab.man.poznan.pl responded with correct results:

73


Fig. 7 MUVI MIB Browser with results.


The MUVI MIB Browser uses the Westhawk's Java SNMP stack. As it is shown above the tested stack supports the IPv6 transport. It was tested on the Linux platform with JAVA 1.4.1. IPv6 requests cannot be performed on Windows platforms, because Windows version of Java Virtual Machine does not support IPv6.

74


5. Summary

In this document 24 network management tools have been identified by the WP6 participant as useful for network management and monitoring in current IPv6 networks.

The following table presents a list of all described tools extended with the status of three requirements: if a tool is IPv6 ready, if it is being used in the 6NET community and if the results of tests have been presented.

Name of the tool IPv6 ready used in 6NET test results provided

1. Analyzer yes no no2. Argus yes yes yes3. ASpath-tree yes yes yes4. CoralReef yes-partially no yes5. Cricket no yes no6. Ethereal yes yes no7. Flow-tools no no no8. JOIN-TV yes yes yes9. IPFlow yes yes no

10. IPm yes no no11. IPv6 Lan Dynamic Topology Discovery yes no no12. IPv6 Looking Glass yes yes no13. IPv6 Management Gateway yes yes yes14. IPv6 support for Netflow v9 in IOS yes yes no15. Mping yes yes yes16. MRTG yes yes yes17. Multicast Beacon yes yes yes18. Nagios yes yes yes19. Netflow/IPFIX yes yes no20. Net-snmp yes no yes21. Network Weathermap yes yes yes22. Pchar yes yes yes23. Polyphemu no no no24. Rancid yes yes yes25. RIPE NCC Test Traffic Server yes yes yes26. SNMP_Session.pm yes no no27. Westhawk's Java SNMP stack yes no no

number of tools for Yes 24 18 14percent of tool for Yes 89% 67% 52%

Out of 27 described tools, 24 (which is 89%) are IPv6 ready. In the area of network management and monitoring, compatibility with the IPv6 protocol is often a complex issue because it requires existence of IPv6 enabled protocols (like i.e. SNMP protocol) for management, whose protocols have to be implemented on monitored network devices and, moreover, the transport for these protocols also has to be implemented. Not all of these requirements are always fulfilled (i.e Management Information Base (MIB) standards for network management protocols are still under

75


76

development). The issues concerning management protocols and MIBs in the standardisation process were covered in D6.3.1 6Net IPv6 Network Management Cookbook.

The selected IPv6 ready tools are mostly identified as useful for IPv6 network monitoring tasks and 18 of them (which is 67%) are being used in 6NET.

Over half of the presented tools – 14 (which is 52%) have been tested by WP6 participants and the results have been presented in this paper.

D6.2.3 shows that there is a set of tools varying in their complexity and functionality, which have been recognized as useful today for IPv6 network monitoring and management. In this paper one can find an appropriate tool for their need together with technical remarks and the results of tests which present their advantages and disadvantages.

In the next deliverable D6.3.2 Implementation of tools and operational procedures, the real use of the tool in the IPv6 test network in the 6net environment will be presented.

6. Bibliography

[1] 6NET Deliverable 6.1: “6NET Management Architecture”, version 1.0, July 2002.

[2] 6NET Deliverable 6.2.1: “6NET Management Tools Requirements”, version 1.0, July 2002.

[3] 6NET Deliverable 6.3.1: “6Net IPv6 Network Management Cookbook”, version 1.0, September 2002.

[4] 6NET Deliverable 6.3.2: “Implementation of tools and operational procedures”, version 1.0, July 2003.

ist project # 2001-32603 - 6net · brief description of the product (brief functionality etc)...

Documents