Pure vs EMC: who’s winning?

by Robin Harris on Tuesday, 11 November, 2014

Forbes contributor and analyst Peter Cohan writes on seemingly conflicting stories coming from Pure and EMC. Let’s unpack the dueling narratives.
Pure logo
Does Pure win 70% vs EMC or does EMC win 95% vs Pure?
The metrics:

Both parties seem to agree that they meet up very often in competitive bidding situations. EMC claims that it beats Pure 95% of the time they vie for “pure flash” bids while Pure says it wins the vast majority of these business battles.
Pure claims that it’s growing at 400% a year and much of that comes from winning 70% of its competitive battles. . . .

The qualifier from Scott Dietzen, Pure CEO:

We beat EMC 70% of the time when we get potential customers to test our product in competition. . . .

EMC spokesman Dave Farmer says:

The fact is, EMC XtremIO wins over Pure 95% of the time in head-to-head all-flash array sales.

Whose numbers can you trust?
Pure’s statement relies on bake-off numbers, where the prospect tests both the products. The hassle involved in getting hardware into a shop for testing means that Pure has a clear record of each competitive test.

EMC’s number is much softer. How does EMC corporate know that they are competing with Pure in a sale? The EMC sales rep says they are.

Sales reps have no incentive to downplay the competitive environment. Indeed, they have a positive incentive to exaggerate the competition since they may need a competitive allowance to close a deal.

The StorageMojo take
Given the apples-to-oranges comparison, both companies can be correct. Pure’s problem is that it takes working capital to put hardware on a customer site and personnel to support the testing.

The good news for Pure is that once they win a sale they have a customer champion in that account. Follow on sales should be easier.

EMC benefits from their large customer footprint. Selling XtremIO is the easy part: their problem is preserving VMAX sales in the face of obviously superior price/performance of flash arrays, even their own.

That’s why the DSSD buy – rumored to cost as much as $1 billion if all milestones are met – is so important. They want to preserve their VMAX software investment while making VMAX performance competitive with flash arrays.

But if customers discover that much of the VMAX software stack is less important in a cloud world even DSSD will be in vain. Which motivates EMC’s effort to also build a comprehensive set of cloud on-ramps – such as TwinStrata – vital.

That is a subject for another post.

Courteous comments welcome, of course. Winning isn’t everything, it’s how you define winning that counts.

{ 17 comments… read them below or add one }

Kevin Stay November 11, 2014 at 2:12 pm

I have no idea who is selling more AFA kit, but I do know if I were in the market it would be no contest; we would be a Pure customer.

Eric Stolch November 11, 2014 at 2:55 pm

Pure ‘s saleforce is incentivized in conducting POCs regardless of the competitor and whether or not they (competitor) have POC gear. So Pure’s data is as valid as EMC’s because both hinge on the rep.

That said, I would claim Pure’s win rate against EMC is not what it used to be and neither are the follow up sales. In fact, most people have bought into flash beyond their true requirements with ample headroom. Furthermore, Pure’s general POC conversion rate has also slowed down given that the larger vendors have taken the necessary steps to more effectively compete in this space.

So bottom line. Neither number is accurate.


Robin Harris November 11, 2014 at 8:25 pm

Eric, so the implication is that if a customer tries Pure, they buy it 70% of the time. And you are correct that Pure is also relying on the rep/SE to report on the competitive landscape. However, it is much more likely that EMC is already in the account than the reverse. EMC doesn’t have an AFA problem; they have a VMAX problem.

OTOH, I’ve rarely met a buyer who was concerned about getting too much performance for their budget. IT loves headroom because running out of resources typically happens at peak demand.


Lou November 13, 2014 at 8:35 pm

I’d have to agree with Robin that there tends not be ‘too much headroom’ especially since it is really easy to fill an AFA from a capacity point of view. So now you have group A on AFA and group B isn’t. Guess who is going to be lobbying for getting on the AFA?

I don’t see what steps the other larger vendors have taken to compete in the AFA space – unless you mean they are now spending even more marketing dollars on lipstick for their current pigs.

Thomas November 15, 2014 at 4:59 am

What about IBM and its FlashSystemss?
According to gartner, they are the one who lead the AFA market…

Reid November 23, 2014 at 12:43 pm

Agreed that EMC has a VMAX problem. One thing your article doesn’t specifically explain is what Pure and EMC are fighting over. Based on what I’ve seen its small-to-medium enterprise replacing the likes of NetApp FAS, HP 3PAR, and EMC VNX. I don’t know a single VMAX customer who would consider Pure or – sadly for EMC – XtremeIO as replacement for a VMAX. I think the AFA players that have the rich data management features that large enterprises require – and there are not many of them – will eventually win that space.

But as you say, if the cloud can replace them, this may all be a moot point.

Chad Sakac November 25, 2014 at 11:55 am

Disclosure – EMCer here…

Robin – my 2 cents… As you say – and I paraphrase – I think both statements can be simultaneously true, and also simultaneously useless 🙂

I can tell you that the 95% number comes from our SFDC records where we compete (directly, or in concert with a channel partner). But – as you point out – that doesn’t reflect campaigns where we don’t even know it’s happening (a competitive displacement). In those case, bad on us as EMC, and bad on the partner that’s supporting the customer for not proactively coming to the customer, and sitting on our laurels.

I would also argue that Pure’s business is already at a sufficient scale that if they think they are aware of every competitive battle something is wrong (at some point, the product/company stands on itself – not just the sales person – that’s true for Pure, and it’s true for EMC).

It’s also no longer purely a subjective thing…. I would point to the #IDC numbers just published (http://idcdocserv.com/252304e_EMC).

Look, I’m not pooh-poohing Pure, they are a respected competitor, and I think Pure is doing just fine.

That said, I think that EMC XtremIO is doing better – and it’s a tricky spot to be the #2 player and growing more slowly than #1. Those graphs don’t look right over time. One can claim “david and goliath” all one wants – in the end, the market speaks.

It’s not just what I think – there is a pile of mounting data (not just IDC) that supports that.

There are a couple of other notables. The IDC analysis doesn’t include Q3, where we really started to hit the gas and accelerate. We reporting publicly on our results as a public company. In Q3, we crossed a $500M run rate with XtremIO (which means we did about the same in Q3 that we did in all of 1H). There’s also some really interesting nuggets. Divide the 1H’14 revenues by raw shipped capacity. Assume we are in striking distance of each other in dedupe/inline (we are) data reduction effectiveness ON AVERAGE (there will always be variation by workload). Wow. XtremIO is a bargain vs. a dual-controller Type I storage architecture.

Furthermore, I think you’re right on another note. EMC doesn’t have a AFA problem. We have other secular trends we must grapple with that are AS disruptive (SDS + COTS, Public Cloud workload migration).

For EMC, there’s a certain amount of XtremIO which is cannibalization – but for a new entrant, every win is net new customers and revenue. That’s the nature of incumbency (has advantages, and disadvantages).

Lastly, I will point out that people are really missing why customers choose HDS or VMAX.

For those that picked those “Enterprise” storage arrays simply for “consistent low latency transactional workloads” – YES, those belong on an AFA **today** (and we would argue that XtremIO is a great choice).

BUT – there are real workloads that need replication and data services that NO AFA on the market offers. I think those are a subset of the workloads currently sitting on that strata of storage architecture (see my cannibalization comment) – but it’s important to note that when those are needed (think hundreds-thousands of devices in consistent sync/low RPO async replication, or massive Oracle environments that depend on things like T10DIF)…. for those, AFA is not a viable choice – at least for now – and those customers are doing Hybrid Flash Arrays configured with dense flash (but that are missing critical AFA data reduction techniques and write amplification reduction approaches).

In any case, while I’m sure some will disagree with my comments, thanks for the dialog Robin!

Vaughn Stewart December 3, 2014 at 2:46 pm

== Disclaimer: Pure Storage Employee ==


As always, nice article. Fair, insightful and thought provoking – which says a lot for as competitive as the storage industry is.

I’d like to offer a broader, alternative perspective, one that I believe supports the growth claims of Pure, EMC and the AFA market. Consider this an attempt to align the conflicting messages from EMC and Pure Storage.

Barring a significant technological advancement, disk-based storage architectures are clearly in the twilight phase of their lifespan. Before someone sarcastically exclaims, “oh yeah and tape is dead” hear me out. Flash is being adopted as a replacement for disk for 5 primary reasons…

1. Cost: Flash is available from a portion of storage vendors at the price of disk

2. Scale: With 100X IOPs flash is better suited to the demands of clusters and unexpected bursts observed with cloud computing

3. Agility: Flash has a significantly higher success rate post deployment than disk. This allows CIOs to advance the business faster.

4. Business Edge: Sub-millisecond latency that produces greater end user satisfaction resulting in greater engagement, larger sales, etc. There’s a mountain of data on this subject.

5. Simplicity: Flash has no nobs to turn. It’s not just simpler than disk; it’s infinitely simpler than all forms of disk including hyper-converged architectures, host-side caches, etc.

While some may find interest in sales processes or compensation plans and others spin implicit but non-existent product positions like locked configuration, destructive scale-up architectures or feature lists – the actual market accelerator for the adoption of flash has been breaking the cost barrier. The AFA vendors who offer flash at or below – yes below – the price of disk are driving the market forward. By no means does ‘cheap flash’ change the market – leaders in the AFA market must meet and exceed what today’s disk arrays provide – but without the element of disruptive economics the ‘flash revolution’ would not have emerged as it clearly did in 2014.

I appreciate your article and all the positions shared in the reader comments. I’d ask all readers to consider flash as much than a faster replacement for disk.

1. Almost every AFA includes data reduction technologies. These technologies directly skew the value to analyst reports of raw TBs shipped. Usable capacity is the new metric. Every vendor should publicly report the savings of their install base – those who don’t should have their technologies reexamined and their sales claims questioned.

2. Data reduction has made cMLC the replacement for 10K & 15K performance disk arrays. Advancements in flash fabrication rates and capacity increase via technologies like TLC & 3D vNAND, we should expect to see capacity oriented storage platforms adopt flash within 2 years. Flash provides a massive relief on data center power, cooling and rack space, which makes it the ideal storage medium for all data that requires being accessed more frequently than data on tape.

3. Lastly, tape is the perfect partner to flash. Tape provides low cost data archival that requires zero-power to store.

While EMC & Pure Storage will battle it out in headlines for the foreseeable future, I don’t think there’s been a more compelling time to cover the storage industry. I’m looking forward to revisiting this post at the end of 2017… I’m sure we’ll all be smiling at the comments shared here.

– cheers,

Curtis Breville January 19, 2015 at 1:25 pm

Another EMC take:

Robin, great topic! I read your articles when I too was an industry analyst for EMA. I’m excited to chime in.

The article is solid in trying to cut through the claims to understand the truth in the market acceptance of a particular product. However, that focus misses the point and is akin to a new tire company claiming a business should swap out its truck’s tires from its current brand: That doesn’t make FedEx or UPS better companies. While Pure focuses on its one product, EMC is involved in the logistics, business practices, efficiency of the entire business, protection of the entire business, and, ultimately, making FedEx or UPS more successful companies overall. It is big picture and a trusted partner at the C-level vs a tactical point-product for the engineers.

Certainly there are going to be new players in the technology world and leap-frogging of feature & functionality after new releases, but the heart of the matter comes down to this: Businesses do NOT want flash…they want faster results. Sure, flash makes faster results possible, but just having flash is not the goal. Understanding the service level agreements, the business initiatives, and where the business sees itself going and how to get there…THAT is what a partner does and EMC along with its Federation partners have hundreds of ways to help businesses meet their goals. Pure has a product. Does it provide disaster avoidance (not disaster recovery) in the manner vPlex offers? Global replication and switching of workloads to different data centers around the world to take advantage of low-cost power consumption? End-to-End converged infrastructures? No, No, and No.

XtremIO is a part of the portfolio of the worlds leading information management company which has trumped the likes of technology powerhouses such as IBM, HP, Oracle, and NetApp in the data management space. Pure? Welcome to the storage product world. In the point-product world, you’ll make a great competitor. In the solution world where disk is only a component of a much bigger picture, what’s the point? I don’t talk product to customers, I talk datacenter-wide solutions and the level of management who cares about solutions isn’t concerned with what new little company’s product does xyz a little better this quarter.

Howard Wolowitz January 21, 2015 at 7:04 pm

Curtis is 100% on point!

Monardo G.A. April 15, 2015 at 6:58 am

I’ve tested both and XtremIO wins ! (30/40% better response time and IOPS). The only issue on XtremIO is the “replica”: at the moment You need something like recover point for this: PURE doesn’t !


Ryan December 4, 2015 at 11:40 am


EMC wins because it owns vmware?

We should all bow down to the supreme leader because EMC knows whats best for my datacenter?

I think the response from Cisco would be, we play with all storage vendors fairly, not we choose EMC because our CEO’s are golf buddies.


Ozzie December 8, 2015 at 10:21 pm

We tried both and purchased pure

It fits out needs better than emc and poc stats demonstrated this.

emc solution also has some key issues around building blocks for growth and also negative dedupe in some situations.

pure may not have the speed of emc but if offers other benefits which site a subset of customers are looking for. In my case space and storage efficiency needs more than speed, pure offered speed as a by product which is the cream on the cake.

Chad April 6, 2016 at 2:49 am

I find it interesting that XtremIO grew as Chad Sakac says, which I do believe, with such as inferior and unstable product at that time, self-admitted by plenty at EMC. Begs the question, why would EMC push their customers towards a platform saturated in bugs with massive scale disruption and data corruption occurring, as we all heard about? How could they push customers needing VMAX’s high-availability into such a platform? Frankly, it shows the true power of EMC Sales and Marketing and not much else. Today, some of the issues have been fixed after 2 years of waiting and that’s good.

But what kind of “partner” (as one poster stated is EMCs strength) puts their partners and customers in these dreadful positions? I’m with neither company, but my experience is with EMC. If my business partners or anyone for that matter treated me this way, I’d turn to something different, and that therein lies the success that the newer companies are having. Flash is fast on any platform, but what can I expect in a year, 3 years, 5 years? Another 3 year refresh quote with an inflated TCO showing obscene support costs?

You don’t have to like these comments, but these are “front line” perspectives that take into account real feedback and not marketing fluff.

Julia December 3, 2016 at 1:24 pm

Really interesting comparison thanks. For further comparisons and reviews from real users of these solutions and many others (with no vendor interference), you can look at IT Central Station (a “Yelp” for tech solutions).

Here’s a direct comparison between EMC and Pure Storage: https://goo.gl/uW68ch. For a full list of all solutions in this category and how real users have ranked them: https://goo.gl/oUbQ08.

Currently #1 is HPE 3PAR with this Senior IT Engineer saying: “The compression it offers is probably the most valuable feature for us. We’re able to reduce 16 terabytes of data by a 2:1 ratio, giving us the ability to put a lot more onto the 3PAR than our previous solution. We looked at EMC, Nimble, and Pure Storage, but 3PAR is the market leader.” To see the rest of what he says, plus explore others in this category click here: https://goo.gl/6bN42I.

Hope this helps.

Laglorden December 16, 2016 at 9:10 am

The Above post is just a search engine optimisation post so Google will point to their dodgy website (it’s posted on several of the websites that comes up). Please don’t click. Please remove (then you can remove this post too)

Julia December 18, 2016 at 4:27 am

You are welcome to your opinion but our community of 100,000+ IT Professionals find us to be a valuable resource and very helpful.

Leave a Comment

{ 1 trackback }

Previous post:

Next post: