EMC/HDS: Fundamental Strategy Difference
by Claus Mikkelsen on Jul 20, 2011
I’m sure many of you have seen EMC’s latest product announcement, the VMAXe. It’s basically a “mini” VMAX which makes the name rather oxymoronic (VMAXmini?), I think. But I’d like to use that announcement to point out a basic difference in the storage strategies of our two companies.
And to do this, I’d like to start with Brian Gallagher’s quote in the announcement, that the VMAXe is “purpose-built”, meaning, obviously, that it is built for a specific purpose. That’s the line that jumped out at me, since it really demonstrates our strategic difference. Why does one have to build a separate product to address a specific need, when one should be capable of filling that “need” in an existing product?
I wrote a magazine article in late 2003 called “Buckets, Pipes, and Pools” giving my vision of where storage was heading, and among other things, I decried the proliferation of all of the “buckets” of storage the industry was cranking out. It seemed to be heading in the wrong direction. The article was translated into French so I’m not sure you can search on it (unless you speak French, of course), but if anyone wants a copy (in English) I can send it to you. But I think EMC has just invented yet another bucket.
Our strategy on the other hand, can be explained in a nutshell: “One platform for all data”. Yes, you can point to our different products and claim otherwise, but we’re definitely on the road to consolidate different platforms, not crank out more buckets (‘scuse me, a purpose-built buckets) of storage. One platform for everything, where data can fluidly and dynamically move up and down ties of storage until it’s been archived. All data. That’s the difference.
It’s clear to me that the VMAXe is targeted at our own VSP. Our VSP can range from a fully popped and smokin’ 2048-drive 2-chassis model with up to 1TB of cache all the way down to a 14U “controller only” version sitting in a 19” rack.
That little controller-only version can virtualize literally petabytes of external storage, providing all that storage with all forms of replication, Dynamic Provisioning and Dynamic Tiering, etc. We decided NOT to call that the VSPe.
But the point is that the VMAXe is trying to fill this gap, but it doesn’t come close and here’s why:
- The entire configuration range of our VSP runs the same microcode so has all the same functionality. The VMAXe runs a different version of Enginuity.
- The upgrade path on the VSP from the smallest to the largest, is seamless. The VMAXe cannot be upgraded to the VMAX which is why it is yet another “purpose-built” bucket.
- The VMAXe does not support z/OS, encryption, SRDF, and certainly still no support for external storage virtualization. VSP supports all of that and more.
So although the VMAXe attempts to fill a big gap in the EMC lineup, it really does not.
I expect to get comments that “it’s cheaper”, but price is settled on the street, and as David Merrill continually points out in his blog, “cost does not equal price”.
C’mon, guys, let’s stop with the proliferation. You’re making storage harder to manage, not easier.
Comments (3 )
Also to just point out, there is VMAX/SE which is a single engine system, once purchased cannot scale up or upgraded to a VMAX.
It seems lately it is more about packaging and the go to market strategy associated with it, than a scalable system.
Cost != price.
Packaging and go to market strategy.
Most observers did not understand EMC’s acquisition of VMware at the time. In 2008, many (including both yourself and Hu) asserted that customers weren’t asking for Flash drives or even automated tiering. Before that, you argued that SATA wasn’t appropriate for Enterprise Storage.
Today VMAX, VNX/VNXe, Centera, Atmos, Data Domain, and Isilon are all #1 in their respective markets. And even NetApp have conceded that one product cannot effectively fulfill all the requirements of the entire storage market space.
Still, some won’t understand.
Hey Claus – good to read your post. I’d be interested in your thoughts on HDS versus NetApp’s approach to “one platform for all data.” Seems that NetApp has done a great job of covering the biggest swath of market with its unified platform and yet it still is forced to fragment its offerings. ONTAP, Bycast, Engenio, Quantum and now Commvault.
I’ve always seen this as a natural evolution of a small player getting big and needing to expand its TAM in places where the product doesn’t scale or doesn’t fit well– i.e. the marketing that got you where you are today shouldn’t stop you from making sound TAM expansion decisions
But I believe if I infer correctly that you’re laying out a different vision where you really do see a single platform as a reality across what is a sizable market scope. Is this a correct interpretation and why will HDS succeed in your view where others have failed?