VhostNet - Revision history

Root: redirect to the current page

2010-09-20T09:21:02Z

redirect to the current page

2010-09-15T13:16:17Z

-netdev type=tap, not -netdev tap

← Older revision		Revision as of 09:16, 15 September 2010
Line 38:		Line 38:
	qemu-system-x86_64 -m 1G disk-c.qcow2 \		qemu-system-x86_64 -m 1G disk-c.qcow2 \
	-net nic,model=virtio,netdev=foo \		-net nic,model=virtio,netdev=foo \
	-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,'''vhost=on'''		-netdev type=tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,'''vhost=on'''

2010-02-09T18:03:20Z

← Older revision		Revision as of 14:03, 9 February 2010
Line 38:		Line 38:
	qemu-system-x86_64 -m 1G disk-c.qcow2 \		qemu-system-x86_64 -m 1G disk-c.qcow2 \
	-net nic,model=virtio,netdev=foo \		-net nic,model=virtio,netdev=foo \
	-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,'''vhost=on'''		-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,'''vhost=on'''

2010-02-09T18:03:07Z

← Older revision		Revision as of 14:03, 9 February 2010
Line 37:		Line 37:

	qemu-system-x86_64 -m 1G disk-c.qcow2 \		qemu-system-x86_64 -m 1G disk-c.qcow2 \
			-net nic,model=virtio,netdev=foo \
	-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,'''vhost=on'''		-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,'''vhost=on'''

2010-02-09T18:01:29Z

← Older revision		Revision as of 14:01, 9 February 2010
Line 34:		Line 34:

	To enable vhost, simply add ",vhost" flag to nic options.		To enable vhost, simply add ",vhost" flag to nic options.
			Example with tap backend:

			qemu-system-x86_64 -m 1G disk-c.qcow2 \
			-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,'''vhost=on'''


			Older (demo) version usage:
	Example with tap backend:		Example with tap backend:

2010-01-07T17:43:59Z

Added IBM's short term plans

← Older revision		Revision as of 13:43, 7 January 2010
Line 93:		Line 93:
	=== vhost-net driver projects ===		=== vhost-net driver projects ===
	* profiling would be very helpful, I have not done any yet.		* profiling would be very helpful, I have not done any yet.
	* merged buffers.		* merged buffers.
	* scalability tuning: figure out the best threading model to use.		* scalability tuning: figure out the best threading model to use.

Line 130:		Line 130:
	* address most vhost qemu TODOs		* address most vhost qemu TODOs
	* get vhost support merged in upstream qemu		* get vhost support merged in upstream qemu

			== Short term plans for IBM(Sridhar Samudrala, David Stevens, Shirley MA) ==
			* Add GSO/checksum offload support to AF_PACKET(raw) sockets.
			* Mergeable RX buffers support in vhost-net.
			* Defer SKB allocation in virtio_net receive path.

2009-12-29T13:59:13Z

← Older revision		Revision as of 09:59, 29 December 2009
Line 66:		Line 66:
	External to system numbers with bridge+tap and 10GE vxge card:		External to system numbers with bridge+tap and 10GE vxge card:
	qemu with bridge+tap, run with: -cpu host,-rdtscp,+x2apic,		qemu with bridge+tap, run with: -cpu host,-rdtscp,+x2apic,
	host+guest 2.6.33-rc2		host+guest 2.6.33-rc2. mtu 1500.

	* netperf TCP_STREAM, default setup, 100 secs run		* netperf TCP_STREAM, default setup, 100 secs run

2009-12-29T13:50:05Z

← Older revision		Revision as of 09:50, 29 December 2009
Line 66:		Line 66:
	External to system numbers with bridge+tap and 10GE vxge card:		External to system numbers with bridge+tap and 10GE vxge card:
	qemu with bridge+tap, run with: -cpu host,-rdtscp,+x2apic,		qemu with bridge+tap, run with: -cpu host,-rdtscp,+x2apic,
	guest 2.6.33-rc2		host+guest 2.6.33-rc2

	* netperf TCP_STREAM, default setup, 100 secs run		* netperf TCP_STREAM, default setup, 100 secs run

2009-12-29T13:31:36Z

add external to system benchmark numbers:

@@ Line 63: / Line 63: @@
 Still tuning performance, especially guest to host.
 Here are some numbers coutesy of Shirley Ma:
 * netperf TCP_STREAM, default setup, 60 secs run

2009-12-09T11:32:10Z

← Older revision		Revision as of 07:32, 9 December 2009
Line 89:		Line 89:
	=== virtio projects ===		=== virtio projects ===
	* improve small packet/large buffer performance: support "reposting" buffers, pool for indirect buffers		* improve small packet/large buffer performance: support "reposting" buffers, pool for indirect buffers
			* guest kernel 2.6.31 seems to work well. Under certain workloads,
			virtio performance has regressed with guest kernels 2.6.32 and up
			(but still better than userspace). A patch has been posted:
			http://www.spinics.net/lists/netdev/msg115292.html

	=== projects involing other kernel components and/or networking stack ===		=== projects involing other kernel components and/or networking stack ===