SCI 2013Aug29 Latest Microsoft ESRP (E2010 & E2013) performance results

In Compellent, Dell, Disk Enclosure, EonStor, ESRP, ESRP v3/Exchange 2010, ESRP v4/Exchange 2013, Eternus JX40, Fujitsu, Hitachi Unified Storage, Hitachi Vantara, HPE, HPE D6000, HUS 110, HUS 130, Infortrend, SC8000 by AdministratorLeave a Comment

This Storage Intelligence (StorInt™) dispatch covers Microsoft Exchange 2010 (E2010) & Exchange 2013 (E2013) Solution Review Program (ESRP) v3 results.[1] Since we last reviewed this 1,000 to 5000-mailbox category, there have been a number of new E2010 submissions from Dell, HDS, HP, and Infortrend. For E2013 there were four submissions in this mailbox range Dell PowerEdge R720xd, HP Proliant SL4540 Gen8, and two Infortrend EonStor DS16F and DS16E submissions. All of our top 10 charts have new rankings with all new submissions in this category. For this analysis we are charting the Exchange 2013 and 2010 results together as there doesn’t seem to be that significant a difference in the I/O performance between the two versions.

Latest ESRP V3.0 results

We start our discussion in Figure 1 with normalized database transfers per second per 1000 mailboxes (1Kmbx), which we calculate from ESRP primary metrics.

Exchange 2010, Exchange 2013 top 10 database transfers per second per 1000 mailboxes for 1001 to 5000 mailbox categoryFigure 1 ESRP v3.0 Top 10 Normalized DB Xfers/Sec/1Kmbx

In Figure 1 there are five new E2010 submissions (#3-5 & #8-9) since our last review which are the HDS HUS 130 (#3), HDS HUS 110 (#4 & #8), the HP D6000 Disk Enclosure (#5) and HP Proliant DL380p (#9). Both HP systems used SAS connected disk drives, and the D6000 used BL640c Gen8 blades to act as Exchange servers talking directly to the storage. The HDS HUS systems were all FC attached storage. Nobody has come close to the #1 HP StorageWorks P2000 G3 SAS in normalized transfers.

Next, in Figure 2 we turn to transaction latencies.

Top E2010 and E2013 database read access latencies, database write access latency and log write access latencyFigure 2 ESRP v3 Top 10 Database Access Latencies

In Figure 2, new E2010 submissions include the HDS HUS 110 (#2 & 4), HDS HUS 130 (#3), Dell Compellent SC8000 (#5 & 7) and the Infortrend EonStor DS S16S-R2250 (#10). We have discussed this previously but it bears repeating, ESRP accepts any submission with an average latency up to 20msec. As such, some vendor’s don’t believe there’s any reason to try to optimize database and log latencies below 20msec. We disagree. It is our opinion that database latency can impact end-user email processing and should be minimized wherever possible.

Nonetheless, there seems to be some reduction of latency that can be had with a lagged copy. For example the Fujitsu ETERNUS JX40 with lagged database copies used SAS connected storage but the other 4 in the top 5 were all FC connected storage. I would have thought that all the top 5 would have been FC storage. In the bottom 5 there are 3-SAS, 1-FC (Dell Compellent) and 1-iSCSI, (HP P2000 G3).

In Figure 3 we discuss Exchange database backup throughput.

Top E2010 & E2013 database backup summed across all servers all database(s)Figure 3 ESRP v3 Top 10 Total database backup throughput

In Figure 3, we see some new E2013 submissions, HP Proliant SL4540 Gen8 (#4) and the Infortrend EonStor DS S16S-R2250 (#8) as well as a slew of new E2010 submissions from Dell (#7), HDS (#1-3) and HP (#5). The E2013 HP Proliant used 30-2TB SAS disks and the E2013 Infortrend EonStor used 10-4TB SAS disks. One would have to conclude that the size of the disks seem to have no bearing on database backup performance. Maybe this is because backup throughput is primarily a sequential activity. Although, when Jetstress measures database backup there is some underlying random database transfer activity but it doesn’t seem to matter that much.

Recall, that we compute this metric as ESRP doesn’t report it directly. Unclear why ESRP only reports backup throughput per database and backup throughput per server but these are highly configuration dependent.  The total database backup throughput metric we use here tries to eliminate these configuration dependencies.

Finally, in Figure 4 we rank the top ten database transfers per spindle.

Top E2010 & E2013 total database transfers per second per disk spindle (drive)Figure 4 ESRP Top 10 Total database backup (MB/sec) for 1K and under mailboxes

Figure 4 has one new E2013 submission, the Infortrend EonStor DS16F-R2653 (#6) and 5 new E2010 submissions from Dell (#2, 9 & 10), HP (#4) and another Infortrend (#8).  There’s a nice mix of 15K (#1&3), 10K (#2,5 & 9) and 7200 RPM disk drives here. Also we have 2-iSCSI (#1&5), 4-SAS (#3,4,6&10) with the rest FC. I would have guessed that the 15KRPM drives with FC interfaces would have performed better on this metric, but I am wrong again. What seems to work well here is to have fewer spindles. None of the top 10 systems here have more than 100 disk drives, and the median number of disk drives was 16. So confounding my expectations, less drives help in this metric. 

Summary

Before our last ESRP review there was a rush of E2010 submissions prior to the changeover to E2013 results. Unsure whether we should continue to chart E2010 and E2013 results together but at this point there weren’t enough to do a Top 10 in any category for E2013. But as we learn more about the IO improvements in E2013 we may elect to separate them out into their own charts.

Reiterating sentiment from previous reports, there are more mysteries in ESRP reports than in any other performance results we examine. Also, ESRP/Jetstress storage system performance seems designed to be difficult to compare but in our view, merit the effort.

Please, feel free to contact us with any constructive ideas on how to improve our ESRP analysis.  We are always open to change our views, especially if the change can lead to a better understanding and comparison of system performance.  In that regard, our contact information can be found in the footer below or on our website at SilvertonConsulting.com.

For a more complete ranking of current top block storage system performance using our ChampionCharts for Email, OLTP and Throughput results, please see our newly updated (May 2019) SAN Storage Buying Guide available for purchase on our website. If your interested in more details on ESRP reports you may also be interested in our recently updated (May 2019) SAN-NAS Buying Guide also available on our website.

~~~~

[This performance dispatch was originally sent out to our newsletter subscribers in August of 2013.  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 storage performance results please consider signing up for our newsletter.]  

Silverton Consulting, Inc. is a U. S. based Storage, Strategy & Systems consulting firm offering products and services to the data storage community. 


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.