[OmniOS-discuss] Pliant/Sandisk SSD ZIL

[Richard Elling]

On Feb 17, 2014, at 2:48 PM, Derek Yarnell <derek at umiacs.umd.edu> wrote:

> Hi,
> 
> So we bought a new Dell R720xd with 2 Dell SLC SSDs which were shipped
> as a Pliant-LB206S-D323-186.31GB via format.

Pliant SSDs (note: Pliant was purchased by Sandisk in 2011) are optimized for
lots of concurrent I/O operations. This is not the kind of workload generated by
the ZIL, which is more contiguous, single thread-like.

>  The bulk of the storage is
> just a 6+2 RaidZ2 with 3TB disks.  When doing a simple untar (40MB/2000
> files) over NFS (synchronous IO) having the mirrored slog of Pliant SSDs
> seems to offer horrendous performance.  On the order of 18-20 OPs per
> second (via zpool iostat -v 1).

Never ever use zpool iostat to measure application performance. zpool iostat
measures workload to the vdevs, showing back-end operations to disk. As such
there is no correlation to client-side operations of any sort, especially writes and
metadata updates. You'll need to go up the stack and see what it is doing. For NFS
I highly recommend nfssvrtop. To see the response time of the Pliant, use "iostat -x"
or, as I prefer, "iostat -zxCn"

Note: response time (measured by iostat -x as a variant of "svc_t") is the critical 
measurement for NFS workloads. Bandwidth is almost meaningless in analyzing
NFS.

>  We have a number of other ZFS builds
> with ZeusRAM slogs and they work very well and while the Pliant may not
> be as good a choice it shouldn't be as bad as this.

It can be as bad as that.

> # zfs get sync,dedup,logbias,compression,checksum zpool0
> NAME    PROPERTY     VALUE          SOURCE
> zpool0  sync         standard       default
> zpool0  dedup        off            default
> zpool0  logbias      latency        default
> zpool0  compression  off            default
> zpool0  checksum     on             default
> 
> Adding a ramdisk slog device makes everything perform as expected.

yep.  A ZeusRAM will also help.
 -- richard

> Setting sync=disabled of course makes things all better.
> 
> # zpool status zpool0
>  pool: zpool0
> state: ONLINE
>  scan: none requested
> config:
> 
>        NAME                       STATE     READ WRITE CKSUM
>        zpool0                     ONLINE       0     0     0
>          raidz2-0                 ONLINE       0     0     0
>            c1t5000C50057EC4B73d0  ONLINE       0     0     0
>            c1t5000C50057EC6EABd0  ONLINE       0     0     0
>            c1t5000C50057EC7CB7d0  ONLINE       0     0     0
>            c1t5000C50057EC8F23d0  ONLINE       0     0     0
>            c1t5000C50057EC998Bd0  ONLINE       0     0     0
>            c1t5000C50057ECB1BBd0  ONLINE       0     0     0
>            c1t5000C50057ECFEE7d0  ONLINE       0     0     0
>            c1t5000C50057ED03F7d0  ONLINE       0     0     0
>        logs
>          mirror-1                 ONLINE       0     0     0
>            c3t5001E820026DA2CAd0  ONLINE       0     0     0
>            c3t5001E820026DA45Ad0  ONLINE       0     0     0
> 
> errors: No known data errors
> 
> 
> -- 
> Derek T. Yarnell
> University of Maryland
> Institute for Advanced Computer Studies
> _______________________________________________
> OmniOS-discuss mailing list
> OmniOS-discuss at lists.omniti.com
> http://lists.omniti.com/mailman/listinfo/omnios-discuss

--

Richard.Elling at RichardElling.com
+1-760-896-4422



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://omniosce.org/ml-archive/attachments/20140217/04688ba6/attachment.html>