EMC and Isilon

by Robin Harris on Thursday, 18 November, 2010

EMC acquires Isilon for $2.25 billion in cash.

The StorageMojo take
In one acquisition EMC gets:

  • A high-growth scale out storage company.
  • A replacement for the aging and unloved Centera/Celerra lines.
  • A worthy competitor to NetApp’s NAS boxes.
  • Scale out storage expertise.
  • And last but not least, a workable substitute for the much delayed Atmos.

2 angles
First, EMC has now replenished its storage product larder. The company is now on course for a decade’s worth of market share leadership.

The DMX is reaching an inflection point to no or negative growth. Big iron storage arrays will never go away – neither have mainframes – but that is not where the future growth lies. EMC has done an artful job of concealing weakness in the DMX business, but expensive new features can only go so far driving DMX revenue growth.

The Isilon product line frees EMC from total reliance on the aging DMX/Clariion RAID paradigm. With added block storage and iSCSI support, the Isilon line is competitive with NetApp.

Make vs buy
Second, there is the larger dynamic of make versus buy. As Atmos demonstrated EMC is not very good at making. Their cutthroat internal politics do not make for an easy long-term investment/development environment.

EMC’s acquisition model
What Tucci has done, and what competitors could learn from, is a simple model for successful large acquisitions. The successful acquisition has several key features:

  • The company has begun successfully selling into the enterprise.
  • The company is the leader in its market niche.
  • The management team is willing to stay for a couple of years or more.
  • The target company’s gross margins must be in line with EMCs business model.
  • EMC’s sales force can leverage the products or mind share.

Given these conditions and the current low cost of capital, EMC can quickly integrate and grow the business. Absent any of these conditions the product would need lengthy reengineering or rebranding before EMC could monetize it.

Hey Joe
Joe Tucci has done EMC proud. He not only pulled EMC out of the dot bomb crater, but he’s positioned it for market share growth in a trying economy driven by long-term global deleveraging.

But his work is not done. Some two thirds of EMCs market cap is tied to its 90% share of VMware. That means the value of what Tucci has done with EMC proper is not yet reflected in their capitalization. VMware’s long-term position in the market is at risk with Microsoft gunning for it with Hyper-V.

Now the company must execute. The new businesses will take 3 to 5 years of engineering, marketing and sales investment to mature.

EMC’s future is not yet secure. But it is hard to see how Tucci could have done better with acquisitions.

Courteous comments welcome, of course. Congrats to Sujal and his team for pulling Isilon back from the brink several years ago to this excellent exit.
Update: The first couple of commenters noted that Centera was not being replaced by Isilon. I updated the comment to Centera/Celerra as Isilon’s architecture can be used for both. I expect some will not agree. End update.

{ 7 comments… read them below or add one }

Jeff Darcy November 18, 2010 at 11:25 am

Huh? Isilon is more directly comparable to Celerra than Centera, and Celerra is still very much alive and kicking vs. NetApp. I have no love for that group, that’s for sure, but credit where credit is due. Isilon is even less of a substitute for Atmos, which can’t be considered “much delayed” while it’s in active deployment by several *very* large partners. There’s more overlap between Centera and Atmos than between either and Isilon, and even then there’s enough differentiation to justify having both.

I do think that the adoption of a scale-out architecture that doesn’t get most of its performance from having another group’s very fast shared storage behind it is a significant and welcome strategic move by EMC. Let’s not obscure that behind spurious fear-mongering about the age of their product line or their ability to innovate.

Chuck Hollis November 18, 2010 at 11:28 am

Hi Robin — thanks for the nod.

While I agree with most of your points, I’d like to offer up a few clarifications if I might?

Both Centera and Atmos use native object storage models, and not more traditional scale-out file systems. We see the use cases for each vs. Isilon as very distinct: different use cases, different platforms.

Some people look at this as a technology platform (somewhat true) but where it really shines is in its specific use case: scale-out NAS for all those high-bandwidth, big sequential file type applications: video, technical computing, HPC, et. al.

Time will tell, but the current view we’ve got is mostly incremental, and not an obvious replacement for anything in the portfolio.

Thanks!

— Chuck

John November 28, 2010 at 8:34 am

Wow, thank you for your pointer to EMC’s VMware market cap ball and chain.

I try not to invest in tech companies i where i think i know anything about the tech. (Rationale: you miss some, but never get political about architecure – bad for work, worse if your net worth is tied in)

Since you mentioned Tucci’s model,

how much of employee comp – and i the EMC guys i respected were actually the hard core sales, once they sussed I wasn’t a prospect (and they got bored with listening to me, same thing!) – hangs on the stock?

i.e. if VMWare goes south, the sales function walks.

only Q is, where do they go to pull down 6 zeros after their number?

Now i’ll take a tangent, please:

The SSD lot are chucking embedded RAID in a sled.

(or approximately threatening to, but the tech is essentially that)

How long ’till I can address sleds as objects on a standard wire?

Just the protocols M’am.

Then effectively goodbye VMotion. Goes to OS level.

I mean, how much does it hurt to be commoditized (drive manufacturing) and how well do we see a reverse (Apple, if you allow software was commoditizing functionality)?

Sorry for bad analogies, but i’m trying broad brush strokes.

cheers,

– j

John November 28, 2010 at 3:06 pm

This one for Chuck, but imaging facebook levels of “likes” and niceness, kay? :-)

and please try to allow my contract says I cannot go blog;

something like 60% of the economics of any big sales oriented org goes on that function.

How exactly does Big Sales Org Guy get away with arguing tech which most often is lost afer the deal is signed, that moment?

I’d like to say this is just another fundamental misorganization of capital, but that’s beyond my pay grade.

How come VMS sussed files as a OS level function, and the rest are cannibalizing their market lead on an a deprecated write function?

(we know why they didn’t go further, but I think MSFT actually does now, skip the 25 years)

Take that down a level:

The OS is supposed to handle integrity etc. No matter if you’re in a shared nothing.

So, you read the calls, make your app aware . ..

But so many people (UNIX,Win) screwed this, Oracle resold the idea.

OH, now there’s your competitor.

And they ship no iron, so they can out-invest you when they’re done messing the lives of the rats who jumped ship.

Heard of “H.A.R.D.”?

Your sales function hasn’t.

And Larry just reminded us all not to LGPL any code: [sic] “I’ll take their/your code, thank you” [/sic]

Not being funny now, but your outfit can pitch as “Whoa, we got a awesome suport team” and leave it a that. So why the posturing?

“Growth”?

– john

John November 28, 2010 at 3:54 pm

I’m going to be personally stupid, but nothing operational, so .. .

– If you want to scale and you believe the tech blurb, you’re done in.
– If you want serious transactions, not a single fool ever trusted either OS FS calls, or a disk.
– People built huge businesses on this misunderstanding, plus Cobb who is notoriously misunderstood.
– You, me, I mean everyone licenses this. But they changed the game from good tech to legal compliance. Which means r/w integrity plus logging.
– Who changes the numbers on the disk game?
– Not software. (tNTFS is good though; ZFS RIP, merge that stack)
– Who could sell that protocol stack? (n.b. has to be in the silicon)
– Moore’s?
– Accept my sillinesss: NT was VMS. Just took 20 years for them to release the APIs, through a layer, which underlying never opened despite promising hints.
– So, you have the tech, but no sale for it, or else the mass market uses in a simplified way you have no VA.
– Sound like big storage? Utilizsed capacity is the bulk quantity.
– You just spent 10 years plus watching your industry go to hell and some sales guys get rich.

Stalled Release,

also, think appliances,

this time with your Uncle Bob screaming how he lost his childhood photos.

To good to be missed.

-j

John November 28, 2010 at 4:41 pm

Chuck,

last word here,

promise!

Why exactly do you want to sell us hardware to fit use situations?

I cannot go buy a bunch of capacity from your company, knowing i can re-purpose than when my app changes. Why?

I mean, what I want is to do this on the fly within MY codebase, and I have PB’s(*) worth I’d prefer to “outsource” to a proper organizaton, yet every time I have to trade POs versus internal op risk.

This does not work how IOS gained NAT.

Cisco lost everyone who gives about PPS or any kind of that. Execs do not ask for line rate. But i know an entrant in routers in one facing niche who has a chance on the real numbers. So a big AS goes through that soon.

Lose the suit mate. Sell me by usable capacity. I know my overheads. aka the money. That’s still huge digits. Plenty on table. You think like a 1992 Apple, taking 65% by design, even if you go bust. Sure, someone can come back in and fix you.

Meanwhile, med. sized customers have very few options if they think past their bonus. This stuff ain’t so hard. you can sell against itertia.

-j

(*) Not seeing much transactions, but otherwise some scale)

Caleb Snow March 30, 2011 at 6:17 am

Robin,

I hear you presenting the DMX line a lot, but what is your take to the VMAX line? I am currently managing about 2 PB with 2 Symm VMAX. 8 Symm DMX (yes, very archaic, even the new ones), 9 Clariion’s, 5 Celerra’s, and a Filer.

Once Unisphere came out for the Clariion/Celellera, the usability went way up in my mind. Adding FAST tiering definitely didn’t hurt either….

Leave a Comment

Previous post:

Next post: