Portal Home > Knowledgebase > Articles Database > LOW performance virtualization


LOW performance virtualization




Posted by hichkas, 12-01-2016, 03:51 PM
Hello we have server which using for virtualization : RAID card: P410i 512 mb cache raid: 10 drive : 4*2TB SAS 7200 centos: 6.x KVM+virtualizor server memory: 64 GB free memory : 2GB vms : 55 we are running very SLOW performance sometimes is there possible to optimize KVM or our RAID

Posted by Srv24x7, 12-02-2016, 01:05 AM
Hi, You are using a RAID card i.e., it is a hardware RAID, so the disk replication is upto the hardware.. I would like to know whether slow performance is seen on the node or on the VMs?

Posted by hichkas, 12-02-2016, 04:06 AM
thank you for reply. our load increase to 50 80 . so our vms and NODE running slowly . also i have turn off the SWAP on main node to increase our performance

Posted by net, 12-02-2016, 06:01 AM
Did you check the health for all the drives? No failing drive?

Posted by HostYourNet, 12-02-2016, 06:05 AM
I would suggest looking at the total system performance aka, CPU, RAM and HD speed at the times your seeing these issues. IF you still cant see a reason look at each VPS on the to see if that the issue been caused. Are you also upto date on the software side for performance improvements and bug fixes.

Posted by hichkas, 12-02-2016, 06:18 AM
yes here details i have check with : DD and hdparm commands they show sometims 2 mb ! 10 mb speed BUT sometimes its normal , show more than 150 mb ( late night and when nobody work with their vms,)

Posted by HostYourNet, 12-02-2016, 06:26 AM
The HD info you pasted doesn't show the S.M.A.R.T status are you able to get that info? If i'm not mistaken doesn't KVM show you the status of the system in webpage? And the VPS on the system each?

Posted by hichkas, 12-02-2016, 06:37 AM
i get the above info directly from raid card P410 . im using KVM+virtualizor on their monitoring system nothing CPU overloading . unfortunately at the moment virtualizor panel unable show DISK I/O USAGE. so always using TOP,ATOP-d, commands to see which VPSID eating DISK I/O i see that but... server should not be overload or disk bottleneck because 2-3 MB vps utilize !

Posted by HostYourNet, 12-02-2016, 06:49 AM
Looking at the Admin demo Virtualizor does not give a system overview just per VPS. TO be able to find out what causing the issue would need to install a 3rd party monitoring system like - http://www.munin-monitoring.org/ If done correctly it will show you past 2 weeks data + an hourly data grap of your system usage every 5 seconds. I use this on my Cpanel server to see its overall status.

Posted by RackRent, 12-03-2016, 07:30 AM
You can set P410 cache values in % for writte and read. For example (50/50, or 30/70). If no battery for P410, by default write cache is dissabled. Enable write back function in virtualization machine. Turned off swap (in some cases) can slower performance.

Posted by hichkas, 12-03-2016, 07:53 AM
one of our server dont have BBU but i have already enable NO BATTERY CACHE here our configs. + write back function on virtualization machine its recommended? Last edited by hichkas; 12-03-2016 at 07:56 AM.

Posted by RockTBN, 12-03-2016, 11:05 AM
With 55 VMs I would suggest you upgrade to SSD and use a newer RAID card. SAS drives with 512MB cache are not enough for that number of VMs.

Posted by Steven, 12-03-2016, 02:11 PM
While inside a virtual machine, run top and see what the 'steal' % looks like. See: http://blog.scoutapp.com/articles/20...you-be-worried

Posted by RackRent, 12-04-2016, 09:06 AM
try to change values (Cache Ratio: 100% Read / 0% Write) into (50/50, or choose biger value according VM demands - more read or more write needed) + write back function on virtual machine 100% OK. cache=writeback (This mode causes qemu-kvm to interact with the disk image file or block device with neither O_DSYNC nor O_DIRECT semantics, so the host page cache is used and writes are reported to the guest as completed when placed in the host page cache, and the normal page cache management will handle commitment to the storage device. Additionally, the guest's virtual storage adapter is informed of the writeback cache, so the guest would be expected to send down flush commands as needed to manage data integrity. Analogous to a raid controller with RAM cache.) I used it and have gain in I/O performance.



Was this answer helpful?

Add to Favourites Add to Favourites    Print this Article Print this Article

Also Read
Brute Force Attacks? (Views: 629)