SCI’s latest Microsoft ESRP performance report as of October 2017

In Enterprise Cloud Platform, ESRP, ESRP v4/Exchange 2013, FlashStation FS3017, Nutanix, SCv3020, Synology by AdministratorLeave a Comment

This Storage Intelligence (StorInt™) briefing covers Microsoft Exchange 2010 and Exchange 2013/2016 Solution Review Program (ESRP) results[1].  There have been no new 1000 and under submissions and as such we cover the over 5000 mailboxes category again as there have been four new submission two from Dell EMC (Dell EMC SCv3020, one with 7.2Krpm and one with 10Krpm drives), a Nutanix Enterprise Cloud Platform and a new vendor offering from Taiwan, the Synology FlashStation FS3017. Once again, many of our top 10 charts have changed and we present some of these below.

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

In Figure 1 we show our Silverton Consulting, Exchange normalized database transfers per second per 1000 mailboxes.

Figure 1 SCI’s ESRP reported top ten database transfers/second normalized per 1000 mailboxes

In Figure 1, the Synology FlashStation FS3017 came in 2nd place in our normalized database transfers per second.

The Synology results are surprising in a number of ways. First, it was done using 24 1.2TB SSDs. Next their ESRP report says it used 1142TB of storage cache but we believe this is a typo and should probably be 1142GB of cache, which still seems a lot for an AFA solution. Also, the SSDs were connected via SATA 3.0Gb/sec, which with NVMe out, is pretty slow. Next, the storage was connected using iSCSI over 2 10GbE ports. Finally, the host Exchange Severs were configured as 24 VMs operating under VMware’s ESXi hypervisor, which is unusual for ESRP runs at this level (20,000 mailboxes).

For our rationale as to why we prefer normalized rather than absolute database transfers per second, please refer to previous ESRP reports.

Next, we report on top ten ESRP database access latencies in Figure 2.

Figure 2 ESRP Top 10 database access latency results

In Figure 2, we can see the Nutanix Enterprise Cloud Platform came in 2nd place with an average database read latency of 1.1 msec. and the Synology FlashStation FS3017 came in 5th place with an average database read latency of 3.1 msec. Although Figure 2 shows all database latencies, submissions are sorted based on database read latencies alone.

The Nutanix Enterprise Cloud Platform 50,000 mailbox solution used 5 (N+1 redundancy) NX-8150 G5 all flash nodes with a total of 120 1.9TB SSDs (24/node). In this configuration, the Exchange servers were also running as 16 VMs across the nodes that support Exchange database and log storage. The interface protocol used between the servers and storage was listed as iSCSI but this was “completely abstracted”, which we take to mean not physical iSCSI but a software iSCSI, using an internal to the server, software loop back which seldom went outside the node unless needing to access external data.

There appears to be some sort of transition at the Synology submission. The results for #1-4 all seem to have better read latencies than write and those after Synology (#5-10) all seem to have worse read latencies than write. As the first 5 (#1-5) are AFA and the following (#6-10) systems are disk only, it all makes sense except for Synology. Here we have an AFA system that does (database) writes faster than (database) reads. It may have something to do with how they use that >1TB of cache (portion used for writes vs. reads) and their destage processing.

Summary

Unfortunately, the Dell EMC SCv3020 submissions didn’t rate on any of our top ten charts. However, the entry of an unknown vendor into ESRP submissions, at this level, is great news, especially when it performed so well.

To have no submissions in the under 1000 mailbox category over the last 9 months is disappointing. This probably says more about the use of DAS or Microsoft Exchange Online (cloud) offerings for these smaller Exchange environments. 

Constructive comments on how to improve our analyses for Microsoft ESRP (Exchange) or any of our performance reports are always welcome. Moreover, if you detect errors in this or any of our other performance reports, please do let us know and we will correct them as soon as possible.

 [This performance dispatch was originally sent out to our newsletter subscribers in October of 2017.  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.]  

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

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

 

[1] ESRP results from https://technet.microsoft.com/en-us/office/dn756396.aspx as of 27Oct2017

Leave a Reply