Wednesday, May 22, 2013

Could IBM buy $RAX ?

It has been an interesting few weeks in the history of cloud.  Dell swooped on Enstratius and then announced it was pulling out of the public cloud battle. This is a pretty bold set of moves by Dell with a strong whiff of realism.  It's bold because it would require someone to face upto the fact that don't have what they need to compete in the infrastructure space and they need to instead play a broker / platform play.  Dell have pockets of strength such as the Dell DCS team but they're unlikely to come off winners against AMZN vs GooG. It has a whiff of realism because to make such a play also means accepting a much smaller role in the infrastructure space in the future and it's unlikely someone who had not realised this was a fight for survival would have made that move.  Tough decision, good call IMHO.

Some of the gloss on OpenStack is starting to crack, as Dana Blackenhorn pointed out in a woefully inaccurate article.  The old myth of AWS being a low margin business - where do people come up with this stuff from? Amazon almost certainly controls its price drops because computing infrastructure demand is elastic but building data centres is constrained by time, resources and capital.  If you drop the price too quickly, you can increase demand beyond your ability to supply which is especially dangerous if you're already on an exponential growth path. However, despite the lamentable state of the post it does pose an interesting question. Would IBM buy $RAX?

In the cloud infrastructure business, IBM is being thoroughly whooped by AMZN and is facing the threat of GooG.  You can spin virtual data centres as private clouds or how private clouds are the thing as much as you want.  The future is a hybrid model of multiple public commodity providers and currently that list is AMZN + AWS clones, GooG, possibly MSFT and maybe some players in China / India (though I expect them to be AWS clones too).

IBM will have massive inertia to this change due to past success of its server and hosting models. That inertia will happily try and hold back IBM from changing, well past the time that its infrastructure business is lost.  But lost it will be and it's already incredibly late in the day to do anything. However, there is a glimmer.

If (and massive if) IBM bought RAX then could this bring them some direct capabilities in the commodity space at scale?  It'll come with some legacy of an older hosting model, it's a minnow compared to AMZN and it comes with baggage such as the differentiation from AMZN nonsense (though it's perfectly true that many in the OpenStack community are trying to make OpenStack have 'behavioural fidelity' with AWS).  But, it could be a good move to bolster OpenStack and establish a competitive market given certain conditions. 

If (HUGE if) IBM bought $RAX, focused its server and hosting business of this space and made it clear that the past was dead, dropped significant yearly investment especially in the first three years - say around $3-5 billion p.a. and nullified AMZN's ecosystem advantage by becoming a better AWS clone through embrace and extend then yes, they could become a player in the infrastructure space despite previously saying it wasn't their focus. They could rally a competitive market around OpenStack and get it up and running fast.

BUT it would mean making all the right moves and that would take strategic wit, a CEO with Captain Albert Ball's courage, able to spend upto $20 billion over three years in the face of strong internal resistance and determination to take the fight to the market. Meg Whitman seems to be trying to shake HP into life maybe Rometty could do the same for IBM because her predecessors have singularly failed to do so.

Would I  personally take the gamble if I was IBM? Yes, but then I would have flooded the market in 2008 and not let myself get in such a position. And that's the problem, given IBM's current positioning and their past timid actions in this space, I find it unlikely that they'll change unless things becomes desperately clear by which time, they may acquire but it'll be too late. 

It's an interesting question but I do see this idea of acquisition as more a flight of fancy. Based upon past actions I don't believe IBM has what it takes. I do expect IBM to focus higher up the stack, to let this infrastructure business slide and my son to grow up in a world where the idea that IBM built infrastructure is a fading memory. 

Maybe Rometty will change that.  Let's hope so, it'll certainly make things more interesting.


--- Final Notes.

1. I'm an advisor to CloudScaling. I think Randy Bias and the crew are spot on in trying to create 'behavioural fidelity' of CloudScaling's OpenStack system with AWS.

2. Whilst I know many engineers involved in OpenStack and I firmly agree with the approach of using open source to create a competitive market, I don't think the execution and the game play has been particularly sound.

3. I have long held the view that there is a transitional role for private clouds that are AWS clones but I expect that to become increasingly niche in the very near future. The key word is 'transitional'. I do hold the view that a public market of AWS clones is possible.

4. Despite the general media interest, OpenStack is not the only contender. I have a stronger preference towards Apache CloudStack. I know Peder Ulander and several of the CloudStack team and I think that Citrix is playing a sound game. I also know many of the people behind Eucalyptus and Open Nebula and they shouldn't be underestimated either.

5. I was an advisor to Enstratius.