Charles Makes a Point (or Two)

I don't plan on making this a rumor mill blog, but this one's pretty good. StreetInsider.com and other financial sites are speculating that if EMC wins the battle for Data Domain (how could they not?), NTAP will go after CommVault.

Now THAT's an interesting move that has a wild chance at success!  (In case you couldn't tell, that was written with a lot of sarcasm).

If I were CommVault folks, I'd be worried right about now.  NTAP has a great track record of buying software companies and turning them into, well, bought companies.  Topio went away, Spinnaker is nowhere to really be seen outside of a roadmap slide that shows the same 'Coming Soon' graphic quarter after quarter, the Decru product seems to show up less and less every time I cross paths with NTAP -- are you detecting a theme here?  

NTAP does some things very well.  They should be rightly credited with driving the unified storage market to the point it is at today, pushing the thinking of the marketplace to places many didn't think it would go.  They have proven that 'good enough' in many situations is really good enough, and unified storage platforms can effectively solve many storage needs in the middle and upper middle marketplace.

They also do some things not so well.  They don't utilize space on the array very efficiently.  Their implementation of Fibre Channel is a little less performant than I'd like it to be at scale.  Their track record of acquiring at least moderately successful software companies and turning them into a game-changing feature or product line is, well, less than stellar.  They tend to get a little too testy when someone questions their technical methodologies.  Nothing horrific.  They aren't a bad company, and their technology is good, just not as good as others, in my opinion.

To me, this acquisition just doesn't make sense.  NTAP doesn't have a history of being able to sell software effectively.  Bringing a company like CommVault into the fold will take superior execution and understanding that the model for selling software can be quite different than that of selling hardware.  It took EMC several tries to grasp this, and it almost didn't work out very well for them (and in some cases it could be debated that they grasped it in time.  Legato, anyone?).  

How do you go from all but acquiring another hardware player that seems a perfect fit to a forced software buy?  You're not going to take CommVault and integrate it into the filers like you could the DD code.  Even if they could make it work, it would suck down so much performance on the array you'd have to buy FOUR of them to get anything accomplished in your environment.  Do you buy them and just leave them alone, letting them do their thing but add to the bottom line?  Maybe, but that doesn't really seem to gain them much.  Or do you swallow them up hoping it makes you more attractive to HP and they then gobble you both up to replace their anemic and ailing EVA line and obtain a modern backup product?  That doesn't seem to be 'in your face' enough for Donatelli to do, but it might just be enough of a poke in Tucci's eye to do it.  It would certainly stir the pot up a little more, and would make some rational sense.  It seems more likely to me that HP will pursue heavy R & D in the LeftHand product and try to turn some of that 'Invent' power loose to turn the storage world upside down.

What a weird day this turned out to be.  

Chuck Hollis did a pretty decent job (in my opinion) of putting a good public face on the EMC Data Domain acquisition on his blog.  Steve Duplessie is missing the boat a little bit (again, my opinion) in his blog entry on the issue, but I do think he gets a couple of things right.

Steve's point about EMC having 82 ways to do dedupe is misguided and feeds into a FUD-Fest for all EMC's competitors in the marketplace.  Now maybe this should be blamed on EMC's marketing group for not delivering a crystal clear message around deduplication, compression, and single instance storage.  I can buy that argument.  But laying this cacauphony of competing dedupe strategies in the marketplace at EMC's feet isn't right or fair.  Every player in the market has some responsibility for that issue. 

I also don't see EMC having another way of deduplicating data as a negative.  I quite frankly get tired of hearing the mantra'd pablem of "EMC has 82 ways of doing <insert functionality here>.  Which one will they try to sell you today?"  That drivel typically comes from the same camp that tries to fix their plumbing problems at home with a hammer and then wonders why the bathroom is full of water.  Get over it.  If you had billions in the bank to buy companies that fit your long-term strategy you'd do the same thing.  Good strategies aren't built around a single way of doing things.  That's what got EMC into so much trouble in the early part of 2000-2001.  Everything was done their way (Symmetrix) or you were an idiot and not worthy to buy their technology.  EMC provides choice to their customers and flexibility in how to design a solution for their most pressing information problems.  Why does everyone beat them up for that?

Anyway, I could go on for a whole post on that topic alone, so let's get back to the topic of the moment.  The acquisition makes perfect sense on two levels:

1. It stops a major competitor in the marketplace from getting better.  So what if this is a defensive move on EMC's part?  Defense wins championships.  Good for Joe and the EMC Board for executing an excellent defensive maneuver.  As a shareholder, I applaud them for not letting NTAP walk in and change the marketplace (again).  Thank you for finally recognizing that while backup as we know it is in a death spiral, the spiral is slow, and EMC needs to do more than try to convince everyone source-based dedupe is the salve that cures all wounds.  Now, go add to the investment and turn the R & D group loose on this code and get it everywhere as fast as possible.  Don't kill the product, don't take a 5 year ramp to understand how to use it, just go get it done!  I want to see products with elements of this code embedded in them within a year.  You brag about all your R & D, now let's see something real come out of it.
2. The code is good.  The device just works, and it isn't because of the hardware.  It's because of DD's code and approach.  Why not get the best code you can for this feature?  EMC knows more about backup than just about anybody in the marketplace, and when they turn this product line loose on their sales force, watch what happens.  It will be nothing short of amazing, if history is any past indication of potential success.  EMC spends a lot of time training their folks on backup, as well they should, and the only real challenge they'll have is getting their faithful followers to stop spouting the 'Data Domain sucks' mantra.

As a reseller of both EMC and Data Domain, this is a winning combination for us.  We no longer have to hear about how bad DD is from EMC, and fend off the lesser (though workable) Quantum solution when we believe target dedupe is called for.  We will get better support from the larger EMC support org, and have everything under one roof.  That helps our business quite a bit.  It takes some of the best minds in backup in the industry and puts them in one house, which for the partner community makes life easier.  Woo Hoo!

That said, here's what I don't like: 

1. EMC paid way too much for the feature, at least on the face of it.  It's going to be tough to recoup this, even with stellar execution.  They should have bought this company at least 2 years ago.   It was a miss on their part to not have done so.  I'm sure if you add in the lost opportunity cost for NTAP, EMC will make out like bandits, but in Joe's stellar track record of acquisitions, I just think this one is at least a partial miss, only because the timing was so late.  But, sometimes you have to overpay and suck it up.  This is probably one of those times.  It just makes me wonder what else is being missed that could change the landscape of the industry?
2. I fear Quantum is dead.  This is where things get a little ugly.  Can Quantum survive on tape alone?  I doubt it.  Their engine has been soundly decried as inadequate by their largest proponent.  That's not good for business.  I like the tape products for what they do, and it will be disappointing to see Quantum flounder further, as they have some really good folks over there who work hard to do what they do.  Perhaps EMC will finally make it official and forgive the $100M they loaned them by just buying them up, finally putting a tape product directly into their catalog.  There are lots of reasons this doesn't make sense, but some that do.  We'll obviously have to wait and see.

At the end of the day, target dedupe is a technology with a very limited lifespan, just as all VTL type products.  CDP technologies, automated information movement platforms, and improved compression techniques that save replication bandwidth costs will make 'backup' as we know it a dead idea in a matter of years.  How many, I can't say.  But if EMC (and others) keep going down the path of newer, tighter, faster compression and dedupe algorithms that make it possible to transmit terabyte data sets across less than OC-level bandwidth in reasonable timeframes, and maybe even T1/T3 types of bandwidth, we won't be having a conversation about backup anymore.  This applies across the board, little AX4/NX4 companies and V-Max companies alike.  They all suffer from the same problems, and these technologies allow for the problem to be solved at both ends of the scale spectrum. 

So get on the pony and ride, EMC.  Ride fast and get this acquisition done and the product into R & D.  Get us new stuff ASAP!