Hi Artem,

On Feb 22, 2010, at 4:13 AM, Artem Trunov wrote:
Have you had a good experience with KVM in the latest SL5.4? Is it 
production-worthy?

Yes, we have had very positive experiences with KVM for production Windows XP 
guests and test Windows 7, Windows Server 2008 R2, SL4, and SL5 guests.  We 
recently deployed our first production server, and are preparing additional 
servers to replace our old VMWare Server 1.0.9 systems.  Of course we 
investigated ESXi, etc. and found KVM to be the best performing (and cheapest) 
solution for us.

1. Boot is only possible from ide bus, and drives are named hda. It seems this 
is a feature of KVM?

IDE is sufficient for our needs.

2. Virtual machine can not use more than 1 host core, even if 2 virtual cpus 
are configured for it.

We haven't noticed this.

3. some times virtual machines lock up and go into a cpu curning loop, 
consuming 100% of a host core. The solution is to kill qemu process.

We haven't seen this either, but have noticed a tendency for Win7 and Server 
R2008 R2 to bluescreen on KVM occasionally (roughly once a week).  We haven't 
had time to investigate what's causing this, but haven't seen it on test 
hardware using the same images.  We have also noticed significant time drift in 
our Windows guests that wasn't seen on VMWare Server, though this has been 
sufficiently corrected using NTP.

4. virt-manager crashes shortly after first start up and few operations. 
Subsequent restart of virt-manager doesn't lead to more crashes..

virt-manager has never crashed for us.

5. It's slow, presumably because of disk operations. I see some 5MB/s max 
read/write. I run a qemu native image format, stored on gpfs.

We've been using the RAW format which performs very well for us.  The qcow 
format was slow (and reportedly buggy).  Our images are stored on an ext3 
filesystem using 15K SAS drives in software RAID5.

My host system has all latest kernels and patches.

Our host system is also running a fully updated x86_64 SL5.4.

I hope this helps,
Devin

------
Devin Bougie
Cornell University
Laboratory for Elementary-Particle Physics
[email protected]<mailto:[email protected]>

Reply via email to