#opnfv-vswitchperf: OPNFV VSPERF
Meeting started by trevor_intel at 15:00:30 UTC
(full logs).
Meeting summary
-
- Trevor Cooper (trevor_intel,
15:00:38)
- Al Morton (acm,
15:06:04)
- Agenda (acm,
15:06:37)
- Danube Plugfest results -Sridhar (acm,
15:07:35)
- ETSI TST - Al (acm,
15:09:46)
- Dashboard - Mars (acm,
15:10:06)
- Dev-update - all (acm,
15:10:25)
- Summit session planning (acm,
15:10:52)
- Euphrates planning (acm,
15:11:05)
- SS - particularly how we fit our
work/collaborate with other projects (acm,
15:13:35)
- SS- visualization + Bitergia (acm,
15:14:14)
- How do we use VSPERF? RH, Intel, others
(acm,
15:14:48)
- Morgan helped, Pod 12 now allowed to post,
etc. (acm,
15:18:32)
- Swagger tool allows JSON description of
results (acm,
15:19:34)
- further work is needed to get this going, need
input on Naming (acm,
15:20:19)
- show anything that works in the future,
preliminary is ok! (acm,
15:21:12)
- thanks Mars!!!! (acm,
15:21:20)
- Marco was tryng to test IP flows with OVS, Src
and Dst IPs (acm,
15:23:29)
- does not work for MoonGen for now, IP addrs gen
is correct, OVS doesn't like it (acm,
15:24:05)
- Bill M has found one issue with MG traffic gen
of small packets, but didin't help this problem (acm,
15:25:09)
- Trex generator (acm,
15:25:31)
- short term goal to use Trex as another
generator - using API to communicate with Server (acm,
15:26:33)
- Sridhar installed on Node 5, able to run 64
byte test with zero loss, not sure config is correct (acm,
15:28:29)
- performance is low compared to other generators
for 64 byte case (acm,
15:28:53)
- low compared to Prox and MoonGen (acm,
15:29:34)
- Deepak and Sridhar are at same site!
(acm,
15:30:46)
- New normative spec on BM performance of NW
infra (trevor_intel,
15:36:18)
- 2 specs approved 1) NFVI path implmentation 2)
Measurements that systems collect (trevor_intel,
15:37:14)
- DIscuss at Summit (new Spec) (acm,
15:40:36)
- Plugfest test results update (trevor_intel, 15:42:10)
- Re-ran MG tests - still shows poorer
performance on OVS with small packet sizes (acm,
15:43:40)
- Bill's mail points to a problem with packet
sizes (now fixed) (acm,
15:53:38)
- current results stored in google drive
(acm,
15:54:23)
- would be interesting to look at b2b testing
long term (acm,
15:54:38)
- Martin says that the b2b is not consistent,
will help harvest the data (acm,
15:55:01)
- Al would like to look at any differences from
Pod 3 to Pod 12 if available (acm,
15:55:40)
- Xavier has tried to fix latency issue - 64 byte
test looked good, in-progress (acm,
15:57:27)
- Xavier assigned an additional cpu for Latency
testing (acm,
15:57:56)
- must be very aware of configuration for BOTH SW
generator and NFVI + vSwitch (acm,
15:58:27)
- relevance to new work item in ETSI NFV
(acm,
15:59:04)
Meeting ended at 16:53:46 UTC
(full logs).
Action items
- (none)
People present (lines said)
- acm (42)
- trevor_intel (5)
- collabot` (3)
Generated by MeetBot 0.1.4.