If I Were in the Driver’s Seat at EMC

If there’s anything to be learned from the ‘All flash’ market space, it is that confusion and complexity won’t be tolerated.

jorgem 150

Just a couple of weeks ago EMC announced the release of DSSD and an all-flash VMAX (VMAX AF). In that same announcement, they also stated they would be “re-architecting” XtremIO to accommodate 3D NAND, 3.8TB SSD:

emc 2016 product lines p3

Source of Revenue Figures: The Register: http://www.theregister.co.uk/2016/02/01/product_blastoff_by_emc/

The first thing that jumps out at me is that EMC now has three “all flash” offerings that can all be positioned in the same space. A very sincere question: How does EMC reconcile that? If there’s anything to be learned from the ‘All flash’ market space, it is that confusion and complexity won’t be tolerated. NetApp, for example, has gone through multiple iterations of flash arrays largely because customers have not been motivated to invest in the conflicting positions NetApp has taken. Introducing a VMAX AF into the fold and positioning it as a “Tier 1/Enterprise” flash array that exists because -- and I quote Jeremy Burton EMC CXO, “There are some things XtremIO just can’t do” -- can be interpreted as that very same contradictory message that NetApp has been plagued by. This strategy’s got me scratching my head!

Source: http://www.emc.com/en-us/index.htm (Keynote Replay, Jeremy Burton introduction of VMAX AF)

I’d like also to point out my (not necessarily HPE’s) own opinion, because I see three things going on with EMC’s announcement last month.

We’ll start here because I think some of you will stop reading after this. We still don’t have Cy15Q4 numbers but for the most part we have EMC’s last three years of FC Block storage revenue:


You can see the declining revenue overall is pretty clear.  Now let’s look at the three systems that make up those numbers:


XtremIO is absolutely growing and now it’s accounting for 30% of EMC’s FC storage revenue.  Two problems:  30% ain’t what it used to be and VMAX is still going to account for 43% of the total.  Let’s put some $$ to those percentages:


Source of Revenue Figures: IDC. All figures are $US/Millions.

We clearly see the growth in XtremIO and, unfortunately for EMC, we can also clearly see the stark decline in revenue overall. And while the growth of XtremIO continues to be positive, the overall numbers lend themselves to the position that EMC might just have a customer base resistant to swapping out their older, monolithic VMAX’s for XtremIO. Releasing an all-flash VMAX makes perfect sense. 40-plus percent of your overall FC attached revenue is a lot of money on the table that they’re not getting to fast enough with XtremIO, and they’re under pressure from all sides.

In my not-so-humble opinion it’s not an indictment of how they feel about XtremIO but rather a possible admission that they’ve conditioned a customer base to expect the “monolithic/modular” approach to be superior and now they have to deliver something to go get that cash. That seems to be the most plausible conclusion given the growth of XtremIO and the potential disruption that multiple flash offerings create.

DSSD and Dell, perfect

The timing of EMC’s acquisition by Dell, coupled with the sales force being compensated on DSSD, could not have been better for them. I expect to see some coupling of Dell Servers and DSSD into at least one solution, if not multiple.  Remember this whole merger thing might well kill off some partnerships when all is said and done. Off the top of my head the XC product line, the Nutanix partnership, comes to mind as something they’re going to have to replace.  But don’t forget Dell customers and partners are big fans of that product.

And now, onto how I’d do things vastly differently if I were in the driver’s seat at EMC… here goes!

Step 1: Eliminate the overlap – give customers simplicity and clarity.
All-flash VNX and VMAX are essentially legacy devices with SSD. That flies in the face of what has been the XtremIO messaging to date. “Built-for-flash” isn’t just EMC’s marketing buzz line, it’s a religious credo. They certainly cannot continue to tout this message after announcing a VMAX AF where literally no changes were made that address the economics or write sensitivity of flash. So does this message fade away now? Is it replaced by this new “NO one-size-fits-all and that’s why we have an entire portfolio” message that EMC used during their recent ‘Quantum Leap’ announcement?

Having a portfolio that addresses all the needs customers may have is great, but that portfolio should be dedicated to addressing flash and non-traditional storage requirements like Cloud or Big Data …. NOT three completely different platforms just for flash. MAKE IT SIMPLE. Eliminate all the overlap.

Step 2: Those who can, do. Those who can’t, add appliances. STOP DOING THAT!
I think this one speaks for itself. Customers are expecting flash platforms to have horsepower for way more than just performance—which is a function of the flash itself and data reduction. This also makes flash efficient & affordable and allows customers to take advantage of the power and cooling savings that flash brings to the table.

No one wants to buy a flash array only to add an appliance on top of it for data services, or worse, to cover the weaknesses. Having the right architecture, and one OS that scales, not only addresses customer needs now, but into the future. This is the direction customers have clearly asked for.

Step 3: Don’t pull 180’s with your messaging—but if it happens, acknowledge it!
Let’s pick on EMC’s “re-architecting XtremIO” statement to make this point. XtremIO only supported eMLC. They pushed hard to make sure everyone was well aware of the differences between eMLC & MLC, and now they’re “re-architecting” XtremIO to accommodate it. Sure eMLC is a better quality SSD but does anyone care? 3D NAND is great, and as the technology evolves it just continues to improve.   EMC is admitting that 3D NAND is important now. That’s only about a year after everyone else. For all the hullabaloo over 3D NAND, it’s still flash and flash is only media—that just happens to have 20X more IO capability. Owning this means having to answer very tough questions around the need to “re-architect” when no one else has had to. Ball = dropped.

EMC is definitely challenged here…

What EMC has never understood is that it’s not about tweaking or twisting. It’s not about marketing or sales messaging. It’s about engineering. It cannot be argued that only one array has been able to pioneer the evolution from disk to flash successfully and that is HPE 3PAR StoreServ. What EMC has never realized in all their useless comparisons to 3PAR is that HPE didn’t take a short-sighted approach focused on disk vs. flash—3PAR was always focused on the IO itself. By building the right architecture focused on addressing the actual IO, you turn the underlying hardware into nothing more than interchangeable parts with added resiliency. That’s why 3PAR is the architecture that all flash arrays are trying to emulate.

It may have taken us years to embrace this fact, but HPE is a company where engineering leads the way. EMC may be known for their world class sales force and they should be. They’re a sales driven organization, whose corporate pyramid has Sales at the top. HPE on the other hand is “The Engineering Company”. We’re engineering led, engineering driven.

There are times in life when the prefix “re” is a very good thing. Re-bate, for example. Re-affirm. Re-assure. Re-liable. All good words. “Re-architecting” my primary flash array? Not so much.

Thankfully there’s an engineering company that never requires anything so “extreme”.

*IDCWorldwide Quarterly Enterprise Storage Systems Tracker.

Want to read more blogs in this series, and learn more about our all-flash 3PAR StoreServ storage systems designed to accelerate business and simplify IT? Please view these additional resources:

Vish Mulchand’s blog
Manish Goel’s blog
And read more about all flash

The brave new world of Windows 10 license activation
View Comments
Join the discussion
Be the first to comment on this article. Our Commenting Policies