[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-users] Runing dbench on virtual domains, incorrect results!

I'm running dbench on Xen-dom0, Xen-domU and a real machine to measure the 
performance loss of virtual machines in comparison withe real one.

The machine has 2 cpus but I have disabled hyperthreading in the bias and all 
my kernels are compiled with SMP disabled. And also its total available RAM 
is 2GB but I have limited available RAM in the kernel command line of all 
machines. SO my real machine has 900MB, dom0 900MB and also my domU has 

Surprisingly, I see that Xen increases the performance! I have repeated my 
tests many times but the result is the same.

Here is my results:
Real machine, unmodified kernel: 237 MB/sec
Xen-dom0 ( no domU created. dom0 is running lonely) : 261 MB/sec
Xen-domU (just 1 doU is running besid dom0) : 284 MB/sec

Seeing better performance on a domU is not so surprising because however Dom0 
has 900MB, and a domU also 900MB and they have some collaborations to access 
the disk and ..., but I have no idea about what I see on dom0. Is this 
because of bad measurement of time in virtual machines?

I appreciate your suggestion  or justification,

Xen-users mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.