abilene resiliency test
DESCRIPTION
Abilene Resiliency Test. John Moore Technical Director Centaur Lab North Carolina ITEC [email protected]. SDSC Kevin Walsh Tom Hutton Abilene NOC Andrew Lee Matt Davy CENIC Erick Sizelove. Spirent Communications Qwest Internet2 Jeff Boote Chris Heermann. Acknowledgements. - PowerPoint PPT PresentationTRANSCRIPT
Abilene Resiliency Test
John MooreTechnical Director
Centaur Lab
North Carolina ITEC
Acknowledgements
• SDSC– Kevin Walsh– Tom Hutton
• Abilene NOC– Andrew Lee– Matt Davy
• CENIC– Erick Sizelove
• Spirent Communications
• Qwest• Internet2
– Jeff Boote– Chris Heermann
Ominous?
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
Motivations
• OC-192 backbone links from Qwest “unprotected”– Previous OC-48 backbone circuits had
SONET protection (sub- 50 ms)
• Ongoing MPLS investigation requires understanding of various protection options available
Test Objectives• Measure convergence time after a link outage
– End-to-end and backbone edge-to-edge– Use longest convenient path
• SDSC to NCSU
• Examine effect of modifying IGP settings– Chose spf delay value as variable
• Examine different MPLS recovery schemes– Soft LSP (loose hop)– Explicit route (strict hops)– Fast re-route
Test Traffic• Spirent Communications AX/4000 tester in
CA-ITEC and NC-ITEC Labs– One IPv4 stream– Five MPLS streams– Unidirectional
• AMI owamp streams between LOSA and WASH– Actually, we have data for any-any Abilene nodes
• IS-IS monitor tool– Not deployed in time
Test Traffic• Owamp traffic
– 64 bytes, 10 pps• Increased rate for test, will leave as new default
– Measure latency and packet loss (indirectly)
• AX/4000 traffic– 1000 bytes frames, 20 Mbps per stream– MPLS tunnel traffic was Ethernet
• Used Juniper L2 circuits to get vlan-tagged traffic into LSP
– Measure packet rate and packet loss
IS-IS Parameters• Link failure hold time
– Delay in declaring link down– Default: zero
• LSP interval– Rate limit for LSP flooding (per interface)– Default: 30 LSPs per second
• SPF delay– Delay in running spf algorithm – Default: 200 ms
• SPF recalculation time– Time required to run spf algorithm– Abilene: usually < 1 ms, some spikes to 2 ms
SNVA
LOSA
STTL
CA-ITEC NC-ITEC
HSTN
DNVR KSCY IPLS
ATLA
CHIN NYCM
WASH
Primary BB Path
Secondary BB Path
NCNICalREN
Primary & Backup BB Paths
IP traffic - spf delay = 200 ms
0
500
1000
1500
2000
2500
3000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Seconds
Packet Rate (pps) Packet Loss (packets)
IP traffic - spf delay = 50 ms
0
500
1000
1500
2000
2500
3000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Seconds
Packet Rate (pps) Packet Loss (packets)
WASH-LOSA - spf delay = 200 ms
32
34
36
38
40
42
44
46
48
50
52
6.5 6.7 6.9 7.1 7.3 7.5 7.7 7.9 8.1 8.3 8.5 8.7 8.9 9.1 9.3 9.5 9.7 9.9 10.1 10.3 10.5
Time (seconds)
Late
ncy (
millisecon
ds)
SNVA
LOSA
STTL
CA-ITEC NC-ITEC
HSTN
DNVR KSCY IPLS
ATLA
CHIN NYCM
WASH
NCNICalREN
Stitched Path
End-to-end Path
MPLS Path Structure
MPLS Paths
Path Type Backup type Backup path
A End-end Soft LSP Via IGP
B End-end Explicit route Explicit
D Stitched Soft LSP Via IGP
E Stitched Explicit route Explicit
F Stitched Fast reroute Pre-computed across failed link
Path B - spf delay = 200 ms
0
500
1000
1500
2000
2500
3000
3500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Seconds
Packet Rate (pps) Packet Loss (packets)
Path B - spf delay = 50 ms
0
500
1000
1500
2000
2500
3000
3500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Seconds
Packet Rate (pps) Packet Loss (packets)
Path D - spf delay = 200 ms
0
500
1000
1500
2000
2500
3000
3500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Seconds
Packet Rate (pps) Packet Loss (packets)
Path D - spf delay = 50 ms
0
500
1000
1500
2000
2500
3000
3500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Seconds
Packet Rate (pps) Packet Loss (packets)
Path E - spf delay = 200 ms.
0
500
1000
1500
2000
2500
3000
3500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Packet Rate (pps) Packet Loss (packets)
Path E - spf delay = 50 ms
0
500
1000
1500
2000
2500
3000
3500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Packet Rate (pps) Packet Loss (packets)
Path F - spf delay = 200 ms.
0
500
1000
1500
2000
2500
3000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Packet Rate (pps) Packet Loss (packets)
Path F - spf delay = 50 ms.
0
500
1000
1500
2000
2500
3000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Packet Rate (pps) Packet Loss (packets)
Observations (so far…)
• Decreasing spf delay seems to make convergence time worse
• MPLS FRR offers a possible alternative way to minimize convergence time
• More analysis needed
• More testing needed