Bug 665299

Note: this can only be safely done after libvirt gains the ability
to control vhost on/off status as requested by bz 643050:
without this the change would be too risky.

Set the dependency appropriately.

Created attachment 490701 [details]
rhel6.1 vhost-net vs virtio-net

Attach the test result of rhel6.1 vhost-net vs viriot-net.
See obvious performance degradation with UDP (R) and TCP (S) under guest <-> host scenario compared with rhel6.1 virtio-net.

Created attachment 493114 [details]
rhel6.1 vhost-net vs virtio-net megabyte/ cpu sheet

1.Test by rate-limiting the stream to within 10% and 30% packet drops with virtio-net and vhost-net respectively ,there are no regression.

Lost packet rate = 10% 

vhost-net:/root/tool/netperf-2.4.5/src/netperf -w 1 -b 78 -H 192.168.0.13 -l 60 -t UDP_STREAM -- -m 1460 
UDP UNIDIRECTIONAL SEND TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.0.13 (192.168.0.13) port 0 AF_INET : interval
Socket  Message  Elapsed      Messages                
Size    Size     Time         Okay Errors   Throughput
bytes   bytes    secs            #      #   10^6bits/sec
124928    1460   60.00      468000      0      91.10
124928           60.00      420001             81.76

virtio-net:  /root/tool/netperf-2.4.5/src/netperf -w 1 -b 78 -H 192.168.0.13 -l 60 -t UDP_STREAM -- -m 1460
/root/tool/netperf-2.4.5/src/netperf -w 1 -b 78 -H 192.168.0.13 -l 60 -t UDP_STREAM -- -m 1460
UDP UNIDIRECTIONAL SEND TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.0.13 (192.168.0.13) port 0 AF_INET : interval
Socket  Message  Elapsed      Messages                
Size    Size     Time         Okay Errors   Throughput
bytes   bytes    secs            #      #   10^6bits/sec
124928    1460   60.00      468000      0      91.10
124928           60.00      420003             81.76

Lost packet rate = 30% 

vhost-net:/root/tool/netperf-2.4.5/src/netperf -w 1 -b 100 -H 192.168.0.13 -l 60 -t UDP_STREAM -- -m 1460 
UDP UNIDIRECTIONAL SEND TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.0.13 (192.168.0.13) port 0 AF_INET : interval
Socket  Message  Elapsed      Messages                
Size    Size     Time         Okay Errors   Throughput
bytes   bytes    secs            #      #   10^6bits/sec

124928    1460   60.00      600000      0     116.80
124928           60.00      420122             81.78


virtio-net:/root/tool/netperf-2.4.5/src/netperf -w 1 -b 100 -H 192.168.0.13 -l 60 -t UDP_STREAM -- -m 1460  
UDP UNIDIRECTIONAL SEND TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.0.13 (192.168.0.13) port 0 AF_INET : interval
Socket  Message  Elapsed      Messages                
Size    Size     Time         Okay Errors   Throughput
bytes   bytes    secs            #      #   10^6bits/sec

124928    1460   60.00      600000      0     116.80
124928           60.00      420029             81.77

2. attach  megabyte/cpu sheet base on comment #8's sheet. except following regesstions,others vhost-net's performance looks good 
     scenarios            Message size       protocols         drop% 
   guest <-> ext guest      > MTU             UDP R      -35.23%~ -29.5% 
   guest <-> ext guest      512               TCP S      -8.94%
   guest <-> ext host       9000 32768        UDP R      -15%,-51.71% 
   guest <-> ext host       256 ~2048         TCP S      -14.01~ -9.94% 
   guest <-> host           all               UDP R      -77.87%~ -5.73%
   guest <-> host          256 ~10834         TCP S      -31.24% ~ -6.15%
   guest <-> host          32,512~2048        TCP R      -5.58% ,44.40%~ -13.48%

(In reply to comment #15)
Hi MST,

i just file a bug 698541 for tracking UDP (R) performance optimization from guest to host .
For other pointed performance issue,what's your comments? as vhost-net have good performance in most of the secarions ,could i verify it as pass or something else?

We added a comment on udp to virt guide.
I think we can close this.

base on comment # 15 #16 #17 ,change bug status to verified.

    Technical note added. If any revisions are required, please edit the "Technical Notes" field
    accordingly. All revisions will be proofread by the Engineering Content Services team.
    
    New Contents:
Cause: vhost-net kernel module was not being loaded automatically, so vhost-net was not being used by default.

Consequence: lower performance than what is possible when using vhost-net.

Fix: removed vhost-net from /etc/modprobe.d/blacklist-kvm.conf

Result: vhost-net is used by default by qemu-kvm.

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2011-0534.html

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2011-0534.html