[pve-devel] new cache benchmark results

Alexandre DERUMIER aderumier at odiso.com
Thu Nov 8 21:58:24 CET 2012


Hi Martin,

#mkfs.xfs /dev/vdb
#mount /dev/vdb /mnt/   (default xfs options)
#fio --filename=/mnt/test1 --rw=randwrite --bs=4k --iodepth=40 --size=10000M --groupe_reporting --name=file1 --ioengine=libaio 


I take the iops result. (this is an average)

I didn't have tested lvm, maybe it could be great to test it over local storage and over iscsi storage.

Alexandre
----- Mail original ----- 

De: "Martin Maurer" <martin at proxmox.com> 
À: pve-devel at pve.proxmox.com 
Envoyé: Jeudi 8 Novembre 2012 21:05:23 
Objet: Re: [pve-devel] new cache benchmark results 

Thx for testing! What command do you run exactly to get this benchmark number? 
I just want to make sure that I can compare your results with mine in the right way. 

Martin 

> -----Original Message----- 
> From: pve-devel-bounces at pve.proxmox.com [mailto:pve-devel- 
> bounces at pve.proxmox.com] On Behalf Of Alexandre DERUMIER 
> Sent: Donnerstag, 08. November 2012 17:09 
> To: pve-devel at pve.proxmox.com 
> Subject: [pve-devel] new cache benchmark results 
> 
> netapp nfs: raw file 
> --------------- 
> 
> default(no drive cache option) : 8000 
> none: 13000 
> writeback = 8000 
> 
> 
> 
> netapp nfs: qcow2 file 
> --------------- 
> 
> default(no drive cache option) : 5000 
> none: 12000 
> writeback = 5000 
> 
> 
> local raid controller (512MB CACHE with battery): raw file 
> ------------------------------------------------- 
> default(no drive cache option): 10000 
> cache=none : 4000 
> cache=writeback : 10000 
> 
> 
> local raid controller (512MB CACHE with battery): qcow2 file 
> ------------------------------------------------- 
> default(no drive cache option): 9000 
> cache=none : 6000 
> cache=writeback : 9000 
> 
> 
> 
> 
> So writeback seem to be faster on local storage, but slower on network 
> storage. 
> 
> Note : with shared storage, with writeback, I generally see big spike (faster 
> than cache=none), but after big slowdown (near zero) during 5-10s. 
> So I think this is when the host need to flush the datas, it add more 
> overhead. (maybe network latency have an impact...) 
> 
> 
> _______________________________________________ 
> pve-devel mailing list 
> pve-devel at pve.proxmox.com 
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 


_______________________________________________ 
pve-devel mailing list 
pve-devel at pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 



More information about the pve-devel mailing list