QTS 4.1.3 - SSD Caching performance issues?

Questions about SNMP, Power, System, Logs, disk, & RAID.
Post Reply
javierin
Getting the hang of things
Posts: 57
Joined: Sat Aug 09, 2014 9:24 pm

QTS 4.1.3 - SSD Caching performance issues?

Post by javierin »

Hi

Has anybody experienced a really high performance hit when using SSD cache (with the option to cache all transfers including sequential ones)? Since I upgraded QTS to 4.1.3, i have had to disable SSD caching for my volumes since transfers were putting the NAS load to 80+. This exact same setup was working like a charm with 4.1.2

My device is a TS853 Pro, with 7x4TB WD Reds in a RAID6 configuration and one SSD for cache. NFS is the most commonly used protocol, although CIFS and iSCSI are used as well. RAM is also not an issue

Code: Select all

              total         used         free       shared      buffers
  Mem:     16338832     16204976       133856            0        36352
 Swap:       530108            0       530108
Total:     16868940     16204976       663964
dmesg does not show anything out of the ordinary, disks are ok and if I just disable SSD cache, everything is smooth again.
User avatar
storageman
Ask me anything
Posts: 5507
Joined: Thu Sep 22, 2011 10:57 pm

Re: QTS 4.1.3 - SSD Caching performance issues?

Post by storageman »

As afar as I understand cache is still read only, so was this hitting reads or writes?
Is the SSD on the approved list?
BTW if they add write caching you may need two.
I would raise a ticket.
javierin
Getting the hang of things
Posts: 57
Joined: Sat Aug 09, 2014 9:24 pm

Re: QTS 4.1.3 - SSD Caching performance issues?

Post by javierin »

This had been working with the exact SSD with 4.1.0, 4.1.1 and 4.1.2. I have detected the problem and am able to reproduce it following this steps:

Enabling SSD cache and checking the (cache big files) checkbox
Starting to write data via NFS
After a few GB, load average would be around 20 and NAS becomes mostly irresponsive.
Disabling SSD cache makes load average start decreasing and after a couple of minutes, it can sustain hundreds of GB of writes without a hiccup.

Already run a full disk block test on all the mechanical drives and SSD drive. All green with 0 errors.

I was wondering if anyone else had experienced this as well with the latest firmware so I created this thread.
Post Reply

Return to “System & Disk Volume Management”