No flash in the pan
If you read the industry press, it would be hard not to notice that all-flash arrays are hot news in the data storage industry right now. A new startup seems to pop up every month or so claiming to be the next big thing. In the heat of the moment, there are even some large vendors making some pretty big mistakes in the rush to get competitive products to market, and there is more FUD, marketechture and product roadmaps than you can poke a stick at.
Throughout all of this, NetApp has been quietly and methodically driving flash innovation into the market, and have accumulated over 200 flash related patents. Some of that innovation, such as our Mercury host-based flash acceleration project that we unveiled in early 2011, ultimately went on to inspire products at companies like VMware and similar Linux-based solutions. Others, like the flash-based PAM-II, later called FlashCache, became the industry benchmark for efficient use of flash technology for hybrid systems following its release in 2010, and also became the foundation for NetApp’s Virtual Storage Tiering in our hybrid FAS Arrays.
But four years is a very long time in the storage industry, and since then, the price of flash has dropped significantly. While it is still an order of magnitude more expensive than SATA drives for the same capacity, when combined with storage efficiency technologies, such as thin provisioning and deduplication (also pioneered by NetApp), the price and speed of an all-flash system often beats an equivalent array made from unoptimised disks.
NetApp is in a fortunate position when it comes to all-flash arrays because we have three very different offerings. First, we released the EF-Series flash array that packs enormous performance and six nines (99.9999%) of availability into a very small 2RU package. These have gone from strength to strength, and there’s more to come with the EF series. We’ve also recently released the first iteration in our purpose-built Flashray which is targeted at solid state technologies that are only just beginning to come to market, and we also have the 8000 series based all-flash FAS, which has turned out to be one of the most surprising entrants in the all-flash array market.
When you look at the press, most of the hype for flash surrounds capacity optimized all-flash arrays. The interesting thing is that when you look at many of these shiny new offerings, there is a lot of similarity to what ONTAP has been doing for many years already, but without the 20 years of corporate development, data management and ecosystem integration. Features like compression, deduplication and flash-wear levelling might be new to them, but they’ve been part of clustered ONTAP since the beginning. Then there are others features, such as application consistent backups, automation, PowerShell integration, non-disruptive upgrades, robust military grade security and data replication, which NetApp customers can access to through the billions of dollars of R&D that has gone into clustered data ONTAP. Many of these features are only thoughts on a long list of roadmap items for these new entrants, some of whom seem to value getting a new product to market quickly more than they value enterprise class quality.
There are some who might point out that ONTAP wasn’t built specifically for Flash, and while that’s true, it’s also true that NetApp wasn’t built specifically for VMware, or for VDI, and yet its architecture made it the best fit for those workloads. In fact, after we released the FAS8000 line, which significantly upgraded the CPU and RAM in our hardware, one of the first things we did was to start testing all-flash configurations. I think we surprised even ourselves at how good they were, so we spent a bit longer making sure they really were as good as the initial testing showed. It turns out that a lot of the speed came not only from the power of the FAS8000 platform, but from CPU and data layout efficiency improvements we’d built into the latest versions of ONTAP. Once we were sure of ourselves, we benchmarked ourselves against the other solutions that were getting so much attention. It didn’t take us long to realise that we had something that was more flexible, faster, and cheaper to own than the new kids on the block. To make it easier for people to scope, buy and install these configurations, we created a series of All-Flash FAS bundles. Each of those bundles addressed different price-performance points in the market, and all of them easily outperformed our major competitors at a lower price point, and usually with a smaller physical footprint as well.
Whether it’s database workloads, VDI, or just a very high performance storage node in a storage cluster used in a cloud deployment, All-Flash FAS combines the world’s best storage efficiency and data management capabilities with the consistent low latency response times of an all-flash configuration. Of course, if you find your workload didn’t really need all that performance, there’s nothing stopping you from moving that workload to a hybrid, or all-disk node in the same cluster, without disruption, or you can replicate that data to a hybrid array for DR, or to a high-density SATA array for efficient long-term backup.
Those are things no other all-flash offering can do, and that demonstrates why the flexibility of a truly unified storage system is something that will continue to hold value for a long time to come.
It’s this quiet, methodical approach that has resulted in NetApp becoming one of the world’s leading vendors of flash technology. I know that we’d rather stay focused on helping our customers achieve their outcomes with our technology, but sometimes it’s nice to look back on your achievements and boast a little, because NetApp has done more with flash than any other array vendor, bar none. By working with our customers and partners we’ve done remarkable things, including:
- Shipping over one hundred petabytes of flash
- Shipping over 30 petabytes in the last two financial quarters alone
- Built and installed over 35,000 hybrid systems
- Sold more than 1500 all-flash systems
But none of this chest beating really matters. In reality it doesn’t really matter whether you sold more than anyone else or if your array is hybrid or all-flash, purpose built. What matters is that you are helping your customers reduce risk, reduce costs and most importantly accelerate their business.
And if our customers are succeeding, then we are succeeding.