SCI’s latest SPC performance report as of August 2016

In $/IOPS, DataCore, EMC, IOPS, LRT, Parallel Server, SANSymphony V10, SPC, SPC-1, VNX, VNX 8000 by Administrator0 Comments

This Storage Intelligence (StorInt™) dispatch covers Storage Performance Council (SPC) results[1]. There have been four new SPC-1/1E and no new SPC-2 submissions since our report last May. The new SPC-1/1E results are for the DataCore Parallel Server & SANSymphony-V 10.0, IBM Storwize V5030 and Lenovo Storage S2200. The disk-only, IBM Storwize submission was run with Version 3 of the SPC-1 benchmark, so for now we are not including it in our other analysis. The other 3 new SPC-1 submissions were all AFAs.

SPC-1 results

We begin our discussion with top ten SPC-1 LRT™ (Least Response Time) results shown in Figure 1.

scispc160828-001Figure 1 Top 10 SPC-1 LRT results

The previously submitted DataCore SANSymphony-V 10.0 is our new #1 with an 80msec LRT,  the newly submitted SANSymphony-V 10.0 came in at #2 and the DataCore Parallel Server is our new #3 with 90msec and 120msec LRT respectively. Unclear why DataCore submitted another SANSymphony-V 10.0 benchmark and also why they inactivated their previous Parallel Server benchmark while adding a new one here. But be that as it may, DataCore currently holds the top 3 best LRT’s in SPC-1.

Both SANSymphony-V submissions are software defined storage solutions with the SCP-1 benchmark and the SSD storage running in the same servers. The Parallel Server is a more normal storage configuration with FC connections to benchmark servers and SAS connections to backend Dell MD1220 SSD JBODs. They had ~17.3TB of SSDs split across the Parallel Server and the Dell JBODs.

Next, we turn to SPC-1 top ten IOPS™ in Figure 2.
scispc160828-002
Figure 2 SPC-1 Top 10 IOPS™

As one can see, the DataCore Parallel Server has exceeded 5M IOPS in SPC-1, which is pretty phenomenal considering the next best contender (Huawei OceanStor) barely reached 3M IOPS. The new DataCore SANSymphony-V 10.0 came in at #6 with ~1.2M IOPS. There are only 3 systems in the IOPS top ten that have any disks at all, the DataCore Parallel Server (#1), the Huawei OceanStor 18800 (#7) and their 6800 v3 (#10). For the Parallel Server there was relatively little disk storage (~2TB disk/~17TB SSD), whereas both OceanStor systems had plentiful disk storage and relatively little SSD storage. The rest of the systems in Figure 2 are AFA.

In Figure 3 we provide the top ten $/IOPS or SPC-1 PricePerformance™.

scispc160828-003Figure 3 SPC-1 Top $/IOPS

We can see two new submissions in Figure 3. At #2 is DataCore Parallel Server, and at #3 is the Lenovo Storage S2200. The Lenovo Storage has dual controllers with 8GB cache total that is SAS connected to the benchmark servers with 4-400GB SSDs. Why the Lenovo controllers would be SAS connected to the hosts is an open question but it’s one of the few SPC-1 submissions having both a SAS connection to (12Gb) hosts and (6Gb) drives.

Finally, we show in Figure 4 our IOPS/GB-NAND chart.

scispc160828-004Figure 4 IOPS/GB-NAND

In Figure 4, the DataCore Parallel Server is doing ~296 IOPS/GB-NAND and generating ~5M IOPS with their ~25.9TB of NAND. The SANSymphony-V coming in at #5, was the older submission. Still, DataCore with a FC attached storage system was able to achieve ~2X the IOPS/GB-NAND of the next best submission (EMC VNX 8000).

We are still trying to determine the best way to depict IOPS/Flash effectivity and have used IOPS/SSD as well in the past. Neither IOPS/GB-NAND or IOPS/SSD seems to tell the whole story as of yet. And with high capacity (15TB-30TB) SSDs coming out this year, it will be very interesting to see how they perform with either metric. Although, barring a miracle, we’re unlikely to see an SPC-1 submission with one of these for a long time.

Significance

Not sure what to do with the SPC-1 Version 3 IBM Storwize V5030 submission. Until I see how the workload has been changed we will leave Version 3 submissions in its own category and not report on them with Version 1 submissions. Version 3 reporting has certainly changed for the better and we love the supplemental information provided with these submissions.

AFA’s continue to dominate SPC-1 submissions. Still it’s surprising to continue to see any top 10 IOPS hybrid submissions but they are slowly being pushed off the list. LRT top 10 has long been ruled by AFA and it’s likely we will never see a disk or hybrid solution make it to the top 10 in LRT ever again.

As always, suggestions on how to improve any of our performance analyses are welcomed…

[Also we offer more block storage performance information plus our OLTP, Email and Throughput ChampionsCharts™ in our recently updated (November 2016) SAN Storage Buying Guide, or for more information on protocol performance results please see our recently updated (January, 2017) 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 August of 2016.  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 U.S.-based Storage, Strategy & Systems consulting firm offering products and services to the data storage community

[1] All SPC results available from http://www.storageperformance.org/home/ as of 24Aug2016

[2] Available at http://silvertonconsulting.com/cms1/product/san-storage-buying-guide/