The IBM DCS3700 Holds Its Own Among the Metrics that Matter in Its Microsoft ESRP 40,000 Mailbox Results

Reading through a 30+ page Microsoft Exchange 2010 Solution Reviewed Program (ESRP) report is, at best, time consuming. Trying to objectively compare it against the results of multiple other ESRP reports so one may implement the right storage solution for their Exchange 2010 deployment is, in a word, daunting. So key to properly evaluating these ESRP reports is to know which of the metrics provide insight into how a storage system may perform in a real world Exchange 2010 deployment.

Microsoft ESRP Reports

Microsoft ESRP reports are where many enterprises turn to identify the most appropriate storage system to use in an Exchange 2010 deployment. This is understandable given that ESRP reports reveal both how a storage system is configured and performs under a specific Exchange 2010 workload.

Since enterprises differ in size, Exchange 2010 testing on storage systems is done based on various mailbox configurations. These include number of mailboxes, the size of each mailbox, and the usage profile of the mailboxes.  To ensure meaningful results, Microsoft developed a standardized methodology that it makes available to Microsoft Certified Storage Partners. These partners leverage this approach to independently test and measure the performance of Microsoft Exchange 2010 on their respective storage systems.

Storage vendors submit their results to Microsoft where they are reviewed by the Microsoft Exchange product group.  Once approved, they are then published on Microsoft’s ESRP TechNet website by vendor and tested mailbox configuration. Configurations are subdivided into sections of 1 – 1000 Mailboxes, 1001 – 5000 Mailboxes and 5000 Mailboxes or greater so enterprises may identify a storage system that best aligns with the anticipated size of their Exchange deployment.

No Light Read

Despite the value that these ESRP reports provide, the challenge that organizations encounter is interpreting their results. Published ESRP results are anything but a light read as they include multiple pages with highly technical information on:

  • Exchange database configurations
  • Host system log files
  • Server and storage configurations
  • Storage system RAID levels

While these and other details are found in the ESRP results of each system, organizations must go through each one to compare and contrast how well different storage systems perform under different Microsoft Exchange 2010 mailbox sizes. In this respect, organizations are on their own when it comes to interpreting each storage system’s ESRP results and how the storage system may perform in their environment.

Key Takeaways and Insights into the Microsoft ESRP Results

To do so, organizations may assume the following is true about each storage system’s ESRP results.

  • First, the storage system is optimized for ESRP testing. The ESRP has become the generally agreed upon methodology for measuring Microsoft Exchange 2010 performance on a specific storage platform. As such, each vendor configures its storage system to perform optimally when it conducts its Exchange 2010 ESRP testing.
  • Second, backup is an equalizer. Every storage system vendor has to run backup jobs on their Exchange 2010 databases and report on the results. The significance of the backup metrics as compared to others provided in the ESRP results is that it provides a raw measurement of sustainable throughput on the storage system. Equally important, every storage system’s ESRP results will contain these metrics on database backup and recovery performance.
  • Third, IOPS and disk latency influence the number of emails that may be processed. The individual and aggregate database I/O statistics for each Exchange server used in the ESRP test are recorded in the published ESRP results. These provide some insight into how quickly the storage system will be able to process emails.  Further, if the overall system takes too long to complete the ESRP test, it does not pass.

This behooves enterprises to examine these specific metrics among the many included in the ESRP results as they provide a fairly quick and easy way to assess the ESRP results from different storage systems. Seven specific metrics to look for in the ESRP results are:

  • MBs transferred/sec
  • Average MB reads/sec
  • Total number of disks tested
  • Database Disks Transfers/sec
  • Average Database Read Latency (ms)
  • Average Database Write Latency (ms)
  • Average Log Disk Write Latency (ms)

Interpreting the ESRP Results

The backup statistics are meaningful since Exchange 2010 backups are essentially performed the same way in all environments and provide a good sense of how well a storage system will perform regardless of the size of the Exchange database. The Average Database and Log Disk Write Latency statistics also merit attention as they indicate how quickly data will be written to the Exchange database and logs files as incoming emails are received.

It is by comparing storage system ESRP results based on testing done with Exchange 2010 mailbox configurations that are comparable to what they plan to deploy in-house that organizations will gain some insight into the performance they should expect it to deliver if implemented.

While there are many ESRP results which one may review, the number of storage vendors that have published ESRP results in the 40,000 Mailbox Exchange 2010 range is fairly finite. This makes it fairly straightforward to look at the metrics of the ESRP results available through the end of Q1 2012 to include the recently published results of the IBM DCS3700. From these results one may draw some general conclusions as to which one might provide the best performance for this size Exchange 2010 mailbox deployment.

Looking first at MBs transferred/sec, most of these storage systems only reported data transfer rates in the 26 – 37 MB/sec range whereas the IBM DCS3700 reported total data transfer rates in the 46 – 67 MB/sec range, nearly double what the others reported.  Looking at the average MB reads/sec achieved during backup, the others reported rates of about 35 MB/sec while the IBM DCS3700 reported a rate of 62.4 MB/sec, again about double of what its competitors reported.

In looking at aggregate Database Disks Transfers/sec as well as Average Database Read and Write Latency (ms) across the different tested solutions, here we see fairly comparable results between the systems. However where the IBM DCS3700 separated itself from the crowd was in its Average Database Write Latency with an average of 2.53 ms. These results were up to 65% better than other storage systems while using fewer spindles to do so. Further, the IBM DCS3700’s Average Log Disk Write Latency was .65 ms which was up to 3X better than other similar storage solutions.

So does the IBM DCS3700 outperform other storage solutions in 40,000 Exchange 2010 deployments? Looking solely at these statistics, the argument can be made that it does.

A better position to take is that the IBM DCS3700 stacks up very well against competitive storage solutions while introducing other features that its competitors in this space do not deliver as well, such as storage density and power efficiency, which it
demonstrated in its ESRP result
s.

Performance is certainly a variable that enterprises need to verify a storage system delivers before they use it to host a 40,000 Exchange 2010 mailbox deployment. However, storage solutions which optimize data center floor and rack space while also delivering needed performance make a strong argument for being given preferential treatment over other solutions.

IBM DCS3700 Holds Its Own among the Metrics that Matter in Its Microsoft ESRP 40,000 Mailbox Results

Few people have time to review one 30+ page ESRP report much less read multiple reports, do a careful analysis of them and then draw a conclusion. However most do have time to examine some key metrics that provide insight into how different storage solutions stack up from a performance standpoint.

Those enterprises with 40,000 Exchange 2010 mailboxes will find only a few published ESRP results from storage system providers in which the IBM DCS3700 more than holds its own in the metrics that matter. So when one factors in its density and power efficiency features, it becomes quite easy to build a compelling business case as to why an enterprise should select the IBM DCS3700 over a competitive storage solution for its Exchange 2010 deployment.

Click Here to Signup for the DCIG Newsletter!

Categories

DCIG Newsletter Signup

Thank you for your interest in DCIG research and analysis.

Please sign up for the free DCIG Newsletter to have new analysis delivered to your inbox each week.