SCI 2010 Oct 25 latest Microsoft ESRP v3 (Exchange 2010) results – 1K to 5K mboxes

In Dell, Emprise 5000, ESRP, ESRP v3/Exchange 2010, HPE, HPE P2000 G3 MSA, P2000 SAS G3, PowerEdge R510, X-IO Storage (nee XIOTech) by Administrator

This Storage Intelligence (StorInt™) dispatch covers recent Microsoft Exchange Solution Review Program (ESRP)[1] V3 results for Exchange 2010. There have been a large number of submissions this past quarter with at least five new ones in the 1K-to-5K mailbox category discussed below.  Future dispatches will report on the 1K-and-under and the over-5K mailbox categories while previous ESRP V2 and ESRP V3 analysis are available on SCI’s website[2].

Latest ESRP V3.0 results

We start our ESRP analysis with Exchange database and log access latency results.  Recall that this chart is sorted by the database read latencies.

SCIESRP101025-001 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

SCIESRP101025-001 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

Figure 1 Top 10 ESRP database-read latencies

For Figure 1, the Xiotech Emprise 5000 and HP StorageWorks 2000sa take top honors with the latest StorageWorks P2000 SAS G3 subsystem coming in a close 3rd.  Note that the #5 result for Xiotech used 500GB 10Krpm drives while their #1 result used 350GB 15Krpm drives.  As discussed previously, the other anomaly with Xiotech’s #1 result was that it provided no database redundancy, probably a rare configuration for most users.

Other odd results on this chart were for the HP Smart Array, and both Dell PowerEdge R510 subsystems (#7, 6 and 10 respectively).  They all had great database and log write latencies but poor read latencies. It may have something to do with the size of their drives 932GB (for HP) and 2TB (for Dell subsystems), drive speed -7200 RPM and the few number of disk spindles (24, 6 and 12 respectively) used in the testing (being able to cache database and log writes but unable to service random database reads).  However, the #1 Xiotech subsystem also had only 20 drives but used 15Krpm drives.

Next we turn to ESRP database transfer counts for the 1K-to-5K mailbox category which span such a wide spectrum that we have normalized results to accesses per 1,000 mailboxes (1Kmbx).

SCIESRP101025-002 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

SCIESRP101025-002 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

Figure 2 Top 10 normalized DB transfers per second

Surprisingly, even normalized results look pretty skewed.  For Figure 2, the #1 result, by a wide margin, was the HP StorageWorks P2000 SAS G3 subsystem.  Unclear just what HP did right here but using SAS probably helped.  Their #3 result was a similar HP P2000 G3 subsystem but was attached to Exchange servers via FC.  Both subsystems used the same 300GB 15Krpm drives, a similar number of drives (90 and 94, for the #1 and #3) respectively), and similar IO intensity (0.18 IO/sec) but attained vastly different results.  The only other significant change (outside of interface used and mailbox counts) was the size of the mailboxes (2.5GB and 1GB respectively). Why the smaller mailbox size would generate such poor results in comparison is perplexing to say the least.

A couple of caveats for normalized results:

  • Normalized results may or may not scale much beyond the reported mailbox counts.  For example, the #1 result only supported 2K mailboxes and may not handle much more than that.
  • Normalized results can be impacted by over provisioning.  For example, the #1 result only used ~53% of its storage for Exchange services allowing it to use more spindles to spread the workload.
SCIESRP101025-003 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

SCIESRP101025-003 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

Figure 3 Top 10 data base transfers per spindle

Speaking of over provisioning, another way to look at Exchange storage performance is to normalize it over the number of disk drives or spindles used for the solution.  Figure 3 shows the total number of (read and write) database operations per second per drive done by each subsystem.  It’s surprising to see a dead heat here for the top spot.  Realize that the HP subsystem (#2) used SAS and the Xiotech used FC attached storage and both subsystems sported 15Krpm drives.

Some caveats for database transfers per spindle results:

  • Drive speed can affect this metric, i.e. faster drives perform better and four of the top six performers used 15Krpm drives.
  • Drive over provisioning usually harms one’s performance on this metric.  However this did not seem to hurt the #1 Xiotech and #2 HP results which used 48% and ~53% of storage for database space.  In contrast, the #3 Xiotech result used ~87% of its disk storage for databases.
SCIESRP101025-004 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

SCIESRP101025-004 (c) 2010 Silverton Consulting, Inc., All Rights Reserved

Figure 4 Top 10 database backup activity in MB/second/database

Finally, Figure 4 shows Exchange 2010 backup throughput rates in MB/sec/database.  Please note that the top four solutions here all were SAS connected storage.  Of the remaining subsystems, there were two more SAS and the remainder, were all FC attached storage.

With Exchange 2010’s mailbox resiliency through DAGs, database backup activity no longer seems important.  Indeed, there was at least one submission that didn’t even report on this metric.  However, there are many valid reasons for database backups and we continue to believe that a need for mailbox backups will persist.  As such, we will continue to report on this metric wherever we deem it valuable to our analysis.

Conclusions

There have been a limited number of ESRP V3 submissions to date – it’s still too early.  Nonetheless, to date there have been just 11 submissions in this 1K-to-5K mailbox category and just over 37 total reports, with more coming out each quarter. For all categories there were 15 new submissions since our last report.  Unfortunately, most of these were in the over 5K mailbox category to be covered in a future dispatch.

Finally, ESRP/Jetstress results seem destined and designed to be difficult to compare but merit the effort.  Thus, we strive to refine our analysis with each report.  As always, feel free to contact us with any ideas on how to improve.  In that regard, our contact information can be found below or on our website at SilvertonConsulting.com.

This performance dispatch was sent out to our newsletter subscribers in October of 2010.  If you would like to receive this information via email please consider signing up for our free monthly newsletter (see subscription request, above right) or subscribe by email and we will send our current issue along with download instructions for this and other reports.  Also, if you need an even more in-depth analysis of SAN storage system features and performance please take the time to examine our SAN Storage Briefing available for purchase from our website.

A PDF version of this can be found at

SCI 2010 Oct 25 Latest Microsoft ESRP v3 (Exchange 2010) results

Silverton Consulting, Inc. is a Storage, Strategy & Systems consulting services company, based in the USA offering products and services to the data storage community


[1] ESRP results from http://technet.microsoft.com/en-us/exchange/ff182054.aspx, as of 25 October 2010

 

[2] All prior SCI ESRP Dispatches can be found at http://silvertonconsulting.com/cms1/news-4/