This post was originally published on this siteI posted a few months back about ESXi queue depth limits and how it affects performance. Just recently, Pure Storage announced ...
It is no surprise that over the last few months, the competition between David (Pure Storage) and Goliath (EMC) has heated up dramatically in the field. Just like David up-ended the balance of power on the battlefield by choosing a better technology (artillery warfare vs. hand-to-hand combat), Pure set out to do something profoundly disruptive to the incumbent storage vendors—offer an all-flash array (AFA) that costs less per GB than a performance disk array, but that was >10X better in performance, power & space efficiency, reliability, and simplicity.
The Pure recipe is working—of the 1000+ arrays that we have shipped to customers, the majority have supplanted a mechanical disk array. And with Pure’s growth the highest in storage history, this year we expect to supplant five times more disk arrays than we did last year.
Given EMC is the legacy mechanical storage market incumbent, it is no surprise that they are our most frequent competitor. In our typical engagement, we find ourselves competing with their flagship VMAX or VNX product lines (btw – ask yourself, if XtremIO is ready for prime time, why is EMC still selling you VMAX?). Once the customer commences a proof of concept (POC) with Pure, then we often see EMC shift to include their new AFA XtremIO in the conversation. But today’s XtremIO product is not quite ready for prime time, given their current lack of essential native features like:
Today, all of the above capabilities are GA for Pure Storage, except for replication, which is in active beta. No doubt these gaps will close over time (although you can be sure Pure is hard at work opening up new ones). But what’s EMC to do over the next 18-24 months we expect it will take to ship and then harden these features?
The obvious tactic is to try to slow down your competitor to buy time. How to do that? Litigation comes to mind as an option ;-). As does sowing “fear uncertainty and doubt” (a.k.a. FUD) about your principle competitor into the market. Unfortunately, we are getting word daily from current and prospective customers that EMC is aggressively ramping-up the FUD assault in the market. We understand that as customers, receiving vendor FUD can be both confusing and annoying. As we have said before competition is good—it drives innovation, customer value, market growth, and makes us better companies. We are frankly grateful to be getting so much competitive attention from EMC. It’s an inevitable byproduct of our success. At Pure we aim to focus on selling our own product and succeeding in the market on the strength of our technology and company, but the FUD has gotten so aggressive that we felt it necessary to respond to a few points to clear-up the market confusion, and hopefully turn the conversation back to what matters.
As such, we’ve recorded a few short videos that respond to each of EMC’s main FUD claims with real data, so that you can form your own conclusions as intelligent customers.
First, we give some high-level context on the FUD wars, and why EMC is engaging in these FUD wars:
Next, we “dive deep” into the four main claims of the EMC FUD:
EMC contends that Pure Storage arrays lose performance while approaching full. Pure Storage presents data from its entire GA installed base to show that this simply isn’t true, and gives some instructions on how to properly perform a “performance when full” test to see for yourself.
EMC contends that Pure Storage’s deduplication isn’t 100% inline, to which we say “guilty as charged”….unlike XtremIO’s simple 4K dedupe, Pure Storage leverages 5 different forms of data reduction, some of which are inline and some of which are post-process. The result is that Pure achieves >2x the data reduction effectiveness of XtremIO (most principally because XtremIO doesn’t offer compression). We present the data, and ask a simple question – would you rather have a flash array which leverages 5 forms of data reduction for the best overall reduction possible, or one with simple 4K dedupe only which misses over 50% of the reduction potential all the time?
EMC contends that XtremIO makes more effective use of flash, and that their eMLC flash is more reliable. We show simple math to prove that Pure Storage arrays are actually 2x more efficient in their use of raw flash than XtremIO, and share our experience with the dramatic reliability gains in MLC flash compared with disk.
EMC contends that their scale-out approach is better, and that Pure can’t achieve 1M IOPS. We dive into the reality of 4K hero benchmarks vs. real-world performance, and show how Pure out-performs XtremIO in real work workloads. We also explain why 4K benchmarks are designed to mislead customers into unrealistic expectations, and why ultimately testing your real applications and data is critical.
Finally, we’ll turn the discussion to five key points that we believe really matter:
Stepping back to the big picture … There’s a reason Pure is the target of this FUD. In our view, it’s the same reason why Pure’s Net Promoter Score (objective measure of customer satisfaction) is >70 while those of our core competitors tend toward the 20’s and 30’s. It’s also the reason that most of the time customers test Pure head-to-head versus our competitors, they choose Pure. And it’s the reason top public market investors like Fidelity, T.Rowe Price, and Tiger Global are betting that Pure is a long-term independent winner in the storage market.
But your best bet is to listen to all of this FUD (both that from Pure and from our competitors), talk directly to selected reference customers and partners about their own experiences, and then test your top two or three choices in your own labs. Use the FUD to craft your PoC test plan, but draw your own conclusions. This is the only way to cut through FUD to reality. Best case, you will find that there is indeed a dramatic gap in capabilities, a gap that enables your business to accomplish dramatically more for less while reducing risk. Worst case, you will get a better discount from your incumbent vendor just for doing the due diligence (of course, keep the runners up on the hook until you get your best deal).
Thanks for doing your homework to get beyond the FUD, and rest assured the Pure team and our partners around the world will continue to do all we can to ensure Pure Storage is the best product and thereby earn your trust and your business.