This Storage Intelligence (StorInt™) dispatch covers Storage Performance Council (SPC) results. There have been only two new SPC-1 v3 submissions, Huawei OceanStor 5500 v5 AFA and NetApp AFF A800. There have been no new SPC-2 submissions. Unfortunately, the Huawei OceanStor 5500 v5 did not make it into any of our top ten charts. A few of our SPC-1 top ten charts have changes.
We begin our discussion with top ten SPC-1 measured IOPS™ performance in Figure 1.
One can see that the NetApp AFF A800 came in at 6thplace with 2.4M IOPS. The NetApp submission was a cluster configuration that included 6 HA pairs (12 storage nodes). All systems in the cluster were AFF A800 storage and each HA pair included 1TB of DRAM and 24 1.92TB NVMe SSDs. Inter-cluster networking was provided by 2 Cisco 3132 32 port 40gb Ethernet switches.
In Figure 2 we present SCI’s economics of IO performance metric, the SCI computed IOPS/$/GB metric.
For IOPS/$/GB metric, the NetApp A800 came in at 4thplace with ~150K IOPS/$/GB of storage. We prefer this metric for economic performance to the SPC-1 $/IOPS as it’s less prone to SSD dominance. Although this less of a problem nowadays as most new submissions are all AFA. That being said we do have 2 hybrid systems in the IOPS/$/GB Top Ten (Huawei OceanStor 6800 v3 & OceanStor 18800).
All of the top 6 storage systems here either NetApp or Huawei clustered storage systems. It seems that clustered storage has an advantage in high IOPS at reasonable $/GB prices.
In Figure 3 we show a bubble chart with LRT vs. IOPS for those systems that cost under $100/GB. The size of the bubble indicates price of the system ($K).
We have truncated the chart in Figure 3 at 1M IOPS just to see more of the disk-only offerings. As one can deduce from Figure 1, there are quite a few AFA-hybrid systems that support over 1M IOPS. However, the other criteria used in Figure 3 is that storage systems cost under $100/GB. This tends to eliminate some of the more expensive storage solutions. Nonetheless, there are at least 14 AFA-hybrid systems, under $100/GB that have achieved over 1M IOPS, two of which are seen at the top of Figure 3.
It’s always interested to us to see that there are plenty of systems capable of high (>500K) IOPS, under $100/GB with sub 0.5msec latencies. Flash has seemed to have eroded the difficulty of achieving high IO performance, at least for the foreseeable future, or until we start viewing sub msec. performance at a much finer (microsecond) granularity.
It’s pretty impressive to see how quickly enterprise class systems have achieved over multi-million SPC-1 IOPS. You would have to reach down to 16thplace in SPC-1 IOPS to see a system drop below 1M. We include the top 30 IOPS in our SAN Storage Buying Guide available on our website (see below).
It’s unfortunate that the new Huawei OceanStor 5500 v5 didn’t crack any of our top ten charts. It is a mid-range class system so it’s a bit of an unfair comparison to pit it against the other enterprise solutions available these days. Nonetheless, it did achieve a 600K IOPS at and SPC-1 LRT of .24 msec. As you may recall from prior SPC performance reports, the #10 ranked LRT subsystem is an older IBM FlashSystem which scored a 0.18 msec. LRT.
As always, suggestions on how to improve any of our performance analyses are welcomed.
[This storage announcement dispatch was originally sent out to our newsletter subscribers in August of 2018. 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. Also we have recently updated (November 2018) our SAN Storage Buying Guide so if you are interested in purchasing block primary storage, please checkout our Buying Guide available for sale 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