SCI 28Jul2014 Latest Exchange Solution Reviewed Program (ESRP) performance results

In Dell, ESRP, ESRP v4/Exchange 2013, Hitachi Unified Storage, Hitachi VSP, HUS VM, IBM, Nimble Storage, Nimble Storage CS-700, PowerEdge R720xd, Storwize V5000 by AdministratorLeave a Comment

This Storage Intelligence (StorInt™) dispatch covers Microsoft Exchange 2010 (E2010) and Exchange 2013 (E2013) Solution Review Program (ESRP) results.[1] Since we last reviewed this over 5000-mailbox category, there have been at least eleven new submissions from Dell (6), EMC (1), HDS (1), IBM (1), NetApp (1) and Nimble Storage (1). As such, just about every one of our top ten charts has changed.

Latest ESRP V3.0 & 4.0 (E2010&E2013) performance

In Figure 1 we show our top ten database read access latency chart for high-end storage.

 

Multiple bar chart showing Top 10 ESRP Exchange 2013 read, write and log database access latencies

Figure 1 SCI’s Top 10 ESRP read latency chart

The two new systems in Figure 1 are the Nimble Storage (#1) hybrid solution and the IBM Storwize V5000 (#7) mid-range storage. Due to Nimble’s use of SSD storage as a flash cache, it should be no surprise that they performed well in read latency and relatively poorly in writes in comparison with all the disk only storage systems in the rest of this chart. Also, as you may recall, Nimble is an iSCSI scale out storage system and in this test, they were using two Nimble Storage CS-700 storage arrays. Note, it’s fairly unusual to have iSCSI systems benchmarking 100,000 mailbox solutions, and it’s also very unusual to have iSCSI solutions have better database read response times than FC.

The IBM Storwize V5000 (#7) configuration was a standard disk-only solution in an IBM Flex System x420 configuration. The fact that it did so well in response time is a quandary. None of the older V5000s or V7000s ESRP runs have ever came close. This was probably due to new microcode but without more detail it’s hard to say anything else.

Next we discuss another of our favorite ESRP metrics, total database backup MB/sec in Figure 2.

Bar chart, top 10 ESRP Exchange 2013 total database aggregate throughput performance resultsFigure 2 Top 10 ESRP total DB backup throughput

Recall, that ESRP normally reports backup throughput for database and for servers but not for storage systems. We compute the total backup throughput for the whole storage system because it presents a better view of storage system capabilities. In Figure 2 the two new systems are the Hitachi Unified Storage (HUS) VM Dynamic Provisioning (#2) and the Dell PowerEdge R720xd (#6). The HUS-VM is based on Hitachi VSP technology but in a more mid-range hardware platform and seems able to perform database backup well. The Dell PowerEdge R720xd was a SAS solution but other than that, it’s unclear why it did so well here, as all the other systems on these top 10 charts are typical mid-range to high-end enterprise storage.

Finally, in Figure 3 we show the top ten log playback ESRP submissions.

Figure 3 Top 10 ESRP log playback storage systemsBar chart, top 10 ESRP Exchange 2013 log playback performance restults

Recall, that log playback is the average time it takes to playback one-1MB E2013/E2010 Log file. Lately, ESRP reports also are listing the average number of log files played as another metric but due to their rarity it’s not something we can report on yet. The new system on the log playback top 10 chart is the Nimble storage system (#5). Once again, the hybrid storage solution showed its worth and no doubt its great read performance helped here as well.

We like the log playback metric because it’s not impacted by the ESRP simulated IO profile parameter, which can range from 0.1 to 0.8 IO/mailbox in this category and as such has an undue impact on DB transfers/second that a system is tasked with. On the other hand, both data base backup (reported on previously) and log playback are measured on a system running at its maximum with a background ESRP database load that doesn’t seem dependent on the simulated IO parameter.

Summary

There were a lot of new submissions since our last report on the over 5000 mailboxes category. It’s also great to see new storage vendors like Nimble, take ESRP on. From our perspective ESRP runs are the closest to a real-life application IO workload that exists today. The fact that Microsoft considers ESRP to not be a storage benchmark doesn’t make it any less useful. It just means there’s more work to do to effectively isolate and compare storage system performance.

Any constructive criticisms on how to improve our analyses for any of our performance reports are always welcome. Moreover, if you detect errors in this or any other of our performance reports please do let us know and we will correct it as soon as possible.

[Also we offer more block storage performance information plus our OLTP, Email and Throughput ChampionsCharts™ charts in our recently updated (August 2018) SAN Storage Buying Guide, or for more information on some select ESRP performance results please see our recently updated (July 2018) SAN-NAS Storage Buying Guide, both of which are available for purchase on our website.]

[This performance dispatch was originally sent out to our newsletter subscribers in July of 2014.  If you would like to receive this information via email please consider signing up for our free monthly newsletter (see subscription request, above right) and we will send our current issue along with download instructions for this and other reports. Dispatches are posted to our website at least a quarter or more after they are sent to our subscribers, so if you are interested in current results please consider signing up for our newsletter.]  

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 28Jul2014

Leave a Reply