lhc p2p experiments. elephant vs turtle flows, do they matter ? various projects history dynes (fdt...
DESCRIPTION
1.Application Aware Add-ons by the ANSE Project: -PhEDEx for CMS -FTS Plugin -GRID FTP Server -dCache Doors - PANDA for ATLAS 2. Network Aware sFlow data from switches to analyze large flows -Metric: (IP/subnet/TCP/Size) over a certain time (say every 15 sec) Detecting large flows and steer them over dedicated paths with guaranteed bandwidthTRANSCRIPT
LHC P2P Experiments
Elephant vs Turtle Flows, do they matter ?
Various Projects History
DYNES (FDT + OSCARS + Dragon / OESS)
OLiMPS (Multipath modules from FloodLight to OpenDayLight)
ANSE (LHC Application framework + SDN + NSI)
1. Application Aware
Add-ons by the ANSE Project:- PhEDEx for CMS
- FTS Plugin- GRID FTP Server- dCache Doors
- PANDA for ATLAS
2. Network Aware
sFlow data from switches to analyze large flows
- Metric: (IP/subnet/TCP/Size) over a certain time (say every 15 sec)
Detecting large flows and steer them over dedicated paths with guaranteed bandwidth
Fat Flow Metrics could be:- Transfer Rate- IP or Subnets- TCP Port
Fat Flow verification can be done by asking applications for their intent (data set size)
Fat-Flow Detection and Traffic Steering
sFlow(Interesting
Flow Analysis)
Check the Transfer Queue
OpenFlow supported
Switch(es) + NSI (if
multiple domains)
Path is local or Remote
(Site subnets + RTT)
Finding an Alternate Route (w/wo QOS)
Fat Flow Demonstration Plans for SC