Difference between revisions of "Talk:VPP"
(→is rsvp-te supported by vpp ??: new section) |
(→problems of vpp performance: new section) |
||
| Line 4: | Line 4: | ||
I want to use VPP as data plane and FRR as Control Plane. also I want to use OSPF-te for traffic engineering. I want to know that does VPP supports RSVP-TE? if not is there any similar protocol that I can use instead of rsvp-te?? | I want to use VPP as data plane and FRR as Control Plane. also I want to use OSPF-te for traffic engineering. I want to know that does VPP supports RSVP-TE? if not is there any similar protocol that I can use instead of rsvp-te?? | ||
| + | |||
| + | == problems of vpp performance == | ||
| + | |||
| + | hi, here I got some troubles while I try to test how good the performance of VPP is, in both ubuntu(18.04 server) and centos(7.2) system. | ||
| + | |||
| + | There are 8 ports each with one network interface card in the machine I use, after dpdk configed and vpp start running, I birdged every interface in vpp'cli and use flowsend tools to give flows to every port. Each flow is 1000M and 128 package lenth. | ||
| + | Then I found only 6 ports' Rx rate is about 600M and other 2 is 0M, and what I expected is: all 8 ports' Rx rates shall be full 1000M. | ||
| + | When I stop and restart giving flows, I found all 8 ports' Rx rate became 0M and would never recovery. But the cli of VPP is not dead. | ||
| + | Then I have to restart VPP to recover the ports. | ||
| + | |||
| + | '''Any idea what I have missed?''' | ||
| + | |||
| + | here is the device'info that I used in experiment: | ||
| + | CPU: Inter(R) Celeron(R) CPU 3855U 1.60GHz | ||
| + | cpu cores :2 | ||
| + | |||
| + | internal mem: 8G | ||
| + | network interface cards: (I350 1000M)*8 | ||
Latest revision as of 09:16, 30 November 2018
I want to catch SIGSEGV signal and display the call stack of programs in VPP framework, how can I do it?
is rsvp-te supported by vpp ??
I want to use VPP as data plane and FRR as Control Plane. also I want to use OSPF-te for traffic engineering. I want to know that does VPP supports RSVP-TE? if not is there any similar protocol that I can use instead of rsvp-te??
problems of vpp performance
hi, here I got some troubles while I try to test how good the performance of VPP is, in both ubuntu(18.04 server) and centos(7.2) system.
There are 8 ports each with one network interface card in the machine I use, after dpdk configed and vpp start running, I birdged every interface in vpp'cli and use flowsend tools to give flows to every port. Each flow is 1000M and 128 package lenth. Then I found only 6 ports' Rx rate is about 600M and other 2 is 0M, and what I expected is: all 8 ports' Rx rates shall be full 1000M. When I stop and restart giving flows, I found all 8 ports' Rx rate became 0M and would never recovery. But the cli of VPP is not dead. Then I have to restart VPP to recover the ports.
Any idea what I have missed?
here is the device'info that I used in experiment: CPU: Inter(R) Celeron(R) CPU 3855U 1.60GHz cpu cores :2
internal mem: 8G network interface cards: (I350 1000M)*8
