i-4 routing scalability

Taekyoung Kwon

Some slides are from Geoff Huston, Michalis Faloutsos, Paul Barford, Jim Kurose, Paul Francis, and Jennifer Rexford

outline

• What is routing?• Current Internet routing

– Focus on BGP

• Routing scalability• A case study in IP routing: ViAggre• What is the design space?

routing

• How do packets get from A to B in the Internet?

A B

Internet

What is routing

One example• Connectionless forwarding

– Each router (switch) makes a LOCAL deci-sion to forward the packet towards B

A B

R1 R4

R2

R3

R6

R7

R5

R8

Routing is…

• How does each router know the cor-rect local forwarding decision for any possible destination address?– Through info of the network topology– This info is maintained by a routing pro-

tocol

• Information– Table size * update rate

Routing taxonomy

• Distributed* vs. centralized• Static vs. dynamic*

– # of hops vs. traffic load

• Intra-domain vs. inter-domain

Goals of internet routing

• Inter-connection• Fault-tolerant• Scalability• performance• ….

Current Internet routing

Internet routing: two levels

• Autonomous system (AS) level– Inter-domain– BGP

• Router level– Intra-domain– RIP, OSPF,…

Internet structure

Original idea

Backbone service provider

Large corporation

Smallcorporation “Consumer ” ISP

“Consumer” ISP“ Consumer” ISP

“Consumer ” ISP

Smallcorporation

Smallcorporation

Smallcorporation

Internet structure

• The reality is…

Source: Arbor Networks* Why peering?

Internet structure• And many tiers

Tier 1 ISP

Tier 1 ISP

Tier 1 ISP

Tier-2 ISPTier-2 ISP

Tier-2 ISP Tier-2 ISP

Tier-2 ISP

localISPlocal

ISPlocalISP

localISP

localISP Tier 3

ISP

localISP

localISP

localISP

Internet routing• Prefix is advertised across ASs

1

2

3

4

5

67

Client

SNU147.46.0.0/16

Path: 6, 5, 4, 3, 2, 1

Inter-AS routing: BGP

• BGP (Border Gateway Protocol): the de facto standard

• BGP provides each AS a means to:1.Obtain subnet reachability information from

neighboring ASs.2.Propagate reachability information to all AS-

internal routers.3.Determine “good” routes to subnets based

on reachability information and policy.• allows subnet to advertise its existence

to rest of Internet: “I am here”

BGP basics• pairs of routers (BGP peers) exchange routing

info over semi-permanent TCP connections: BGP sessions

• when AS1 advertises a prefix of AS2 to AS3:– AS1 promises it will forward datagrams towards

that prefix.– AS1 can aggregate prefixes in its advertisement

3b

1d

3a

1c2aAS3

AS1

AS21a

2c

2b

1b

3ceBGP session

iBGP session

2.3.4.0/24

2.0.0.0/8

2.3.0.0/16

Distributing reachability info

• using eBGP session between 3a and 1c, AS3 sends prefix reachability info to AS1.– 1c can then use iBGP do distribute new pre-

fix info to all routers in AS1– 1b can then re-advertise new reachability

info to AS2 over 1b-to-2a eBGP session• when router learns of new prefix, it cre-

ates entry for prefix in its forwarding ta-ble.

3b

1d

3a

1c2aAS3

AS1

AS21a

2c

2b

1b

3ceBGP session

iBGP session

Path attributes & BGP routes• advertised prefix includes BGP attributes.

– prefix + attributes = “route”• two important attributes:

– AS-PATH: contains ASs through which prefix advertise-ment has passed: e.g. AS 6431, AS 7018

– NEXT-HOP: indicates specific internal-AS router to next-hop AS. (may be multiple links from current AS to next-hop-AS)

• when gateway router receives route advertise-ment, uses import policy to accept/decline.

Network Layer 4-17

BGP route selection

• router may learn about more than 1 route to some prefix. Router must se-lect a route.

• elimination rules:1. local preference value attribute: policy de-

cision2. shortest AS-PATH 3. closest NEXT-HOP router: hot potato rout-

ing4. additional criteria

Network Layer 4-18

BGP messages

• BGP messages exchanged using TCP.• BGP messages:

– OPEN: opens TCP connection to peer and au-thenticates sender

– UPDATE: advertises new path (or withdraws old)

– KEEPALIVE keeps connection alive in absence of UPDATES; also ACKs OPEN request

– NOTIFICATION: reports errors in previous msg; also used to close connection

Network Layer 4-19

BGP routing policy (1/2)

A,B,C are provider networks X,W,Y are customers (of provider networks) X is dual-homed: attached to two networks

X does not want to route from B via X to C .. so X will not advertise to B a route to C

A

B

C

W X

Y

legend:

customer network:

provider network

Network Layer 4-20

BGP routing policy (2/2)

A advertises path AW to B B advertises path BAW to X Should B advertise path BAW to C?

No way! B gets no “revenue” for routing CBAW since neither W nor C are B’s customers

B wants to force C to route to w via A B wants to route only to/from its customers!

A

B

C

W X

Y

legend:

customer network:

provider network

Network Layer 4-21

Why Intra- and Inter-AS routing different? Policy: • Inter-AS: admin wants control over how its traffic

routed, who routes through its net. • Intra-AS: single admin, so no policy decisions

needed

Scale:• hierarchical routing saves table size, reduced up-

date trafficPerformance: • Intra-AS: can focus on performance• Inter-AS: policy may dominate over performance

Routing table (RT) growth• Multi-homing• Traffic engineering• Non-aggregatable prefix allocation

Routing scalability

routing message updates

• BGP update messages

Why routing scalability mat-ters?

• FIB is expensive

ViAggre

Virtual aggregation (ViAg-gre)

ViAggre: Basic Idea

ViAggre: Basic Idea

ViAggre: Control Plane

More practically,…

Data plane operations

Route stretch

Ingress -> aggregation point

Aggregation point -> egress (1/3)

Aggregation point -> egress (2/3)

Aggregation point -> egress (3/3)

now

• We will consider general routing de-sign space– IP is just one of the possibilities– But IP networking environments had

better be considered as much as possi-ble

Design Space

Design goal of routing

1. Scalability (memory): e.g. sublinear RT size scaling2. Quality (stretch): the length of a chosen path by a

routing scheme compared to shortest path3. Reliability: fast convergence upon topology changes

while minimizing communication costs to maintain coherent non-local knowledge about network topol-ogy

4. Name-independent routing: accommodate node addresses/labels assigned independently of the topology (otherwise need to split locator and ID parts in addressing architecture)

5. Message overhead

Issue 1: Addressing and routing• Rekhter’s Law: “Addressing can follow

topology or topology can follow address-ing. Choose one.”

00 01 02 03

10 11 12 13

202221 23

3031 32 33

2 10 5 15

613 8 1

316 12 9

14

11 4 7

Name-dependent routing Name-independent routing

Issue 2: state vs. stretch

routingdebate

We want small state!!

We want small stretch!!

• State: the routing table size describing the network topology

• Stretch:

path length found by the routing algorithm

optimal path length≥1

More general trade-off

Triangle of trade-offs:• Adaptation costs = convergence measures

(e.g. number of messages per topology change)

• Memory space = routing table size• Stretch = path length inflation