Content Copyright © 2015 Bloor. All Rights Reserved.
Also posted on: The IM Blog
When I, from time to time, mention Cray in conversation with other analysts or users, the usual question I get is “oh, are they still in business”? Well, yes. And growing rapidly. In 2011 company revenues were $236m, official guidelines (they are publicly listed) is that they will turnover $715m this year. That’s some serious growth so the answer is: “yes, they are very much in business”.
My main interest with respect to Cray is in its Urika-GD and Urika-XA products, where GD stands for graph database (the previous YarcData product) and XA stands for extreme analytics, which is a Hadoop appliance. I have written about both of these previously. But of course Cray is best known for high performance computing.
Recently, Cray opened a new EMEA headquarters in Bristol and I attended the opening ceremony. Many of the high and mighty from Bristol and the South West were there as well as users (in particular the Met Office) and one government minister (Ed Vaisey). One of the reasons for locating in Bristol was Cray’s acquisition, a couple of years ago, of the intellectual property and many of the staff of Gnodal, a Bristol based company researching high performance interconnects.
Anyway, that’s all well and good but most interesting was Cray’s assertion that it sees a coming convergence between high performance computing and big data. If this is true then Cray will be very well placed and we can expect continued stellar growth rates. So, is it true?
The short answer is that it does seem likely. It’s not so much big data per se but the Internet of Things driving what perhaps would be better termed huge data. And if you have huge amounts of data that you need to analyse, data in anything approaching a reasonable timeframe, then you are going to need lots of computing power. So, the assertion is probably true.
The key question is whether Cray can bring its expertise to bear in the general enterprise market. It is fine being a leading provider of the very largest supercomputers but most organisations, even with huge data, aren’t going to require the sort of power that the Met Office, for example, does. Cray does seem to appreciate this, at least in theory, and it has organised its sales and marketing efforts by industry vertical. Nevertheless there is a difference between the sale of tin (even if it is commodity-based tin: Intel, Nvidia, Linux and so forth) and solution sales, which is the category that the Urika products fall into. As it happens there are some applications within the high performance category – for example, seismic processing in the energy sector – but graph databases and Hadoop don’t quite fit the same model, so Cray needs to be careful about how it organises and trains its salesforce – when you are selling a database (graph or Hadoop) users may care about performance but they don’t care about high performance computing per se – it’s a different conversation.
In summary, I think the company has some real opportunities and some of its future plans (which I can’t divulge) will increase its presence in the market even further. But if Cray is right about the convergence of high performance computing and big data then it may have to rethink “The Supercomputer Company” tagline – because it will become ubiquitous – just as the market as a whole should drop “big data” as being completely meaningless.