Oracle: What now for HPC?
Groping in the dark after Sun deal done
There hasn’t been a lot of talk yet from Oracle about their plans for HPC now that the Sun purchase has been consummated - so what's going on?
From what I can tell, it’s not that Oracle is necessarily holding anything back; it’s more that its approach to the market is still being shaped up. The big job was getting the systems roadmaps in place and communicated to the world – and there still are some who will argue that Oracle hasn’t been clear enough on that issue.
However, it looks like almost all HPC-related content has disappeared from the Oracle website. If you drill down into the typical HPC market categories – Education & Research, Public Sector, and Life Sciences, for example – you’ll see that Oracle is primarily talking about how these segments are using Oracle software to manage the business side of the organization: human resources, ERP, etc.
It gets a bit deeper in Financial Services, where they cite a number of financial analytic packages. But what you don’t see is any obvious reference to Sun’s HPC organization or any materials about its value proposition or products for the large-scale, high throughput computing market.
For my part, I’ve been through a few of these vendor mergers and have a pretty good feel for how these things shake out. It generally takes longer than anticipated to figure out what will be on the final menu of offerings and which side dishes will be complementing them.
At this point, HPC-wise, it’s still anyone’s guess exactly what they’ll do. But the anecdotal info we’ve gleaned, coupled with some educated guesses and rank speculation, might let us take a stab at describing the probable course.
I think that one thing is very clear: Oracle is all about making money with this acquisition. Sun isn’t a hood ornament or decorative tail pipe; it’s supposed to be an integral part of how Oracle addresses the market and will be expected to contribute both revenue and profits. Oracle has fully bought into the idea that delivering integrated solutions is the best way to exploit the value of what they can now bring to the table. It sees this route as the best way to provide customer value while maximizing its own revenue and profits.
Oracle also doesn’t like to lose. OK, it isn't unique in that respect; no one likes losing. But Oracle really doesn’t like to lose, whether it be in a product evaluation or in a deal. There’s very little sense of “Well, you can’t win ‘em all” or “At least we gave it a good shot”. It isn't wild about participating in any game where it can’t be number one. In the Sun deal, the best example of this is its explicit rejection of the x86 commodity server market.
So what can we draw from this that can help us figure out what Oracle might or might not be doing in HPC? Is it still going to play in the market? Or abandon it?
I think it’s safe to say that Oracle won’t be interested in chasing top 10- or even top 50-sized supercomputer deals. These deals are notorious for requiring almost superhuman effort on the part of the vendor to bring them home.
Win one of these deals and you can generate a lot of good publicity, provide a halo over the rest of your product line, and even learn a lot about how to build massive systems and make them work. The downside is that you probably won’t make any money on it. In fact, you’ll probably lose money at the end of the day – maybe a huge amount if there are problems. (And there are almost always problems.)
Some vendors need these deals as proof points for their technology and ability to handle the pressure of a huge and complex installation (think SGI, Cray, others). Other vendors quite happily eschew this part of the market and make bank on smaller and much more profitable wins (HP, Dell, others). I think Oracle will fall into this latter category.
IBM has a foot in both camps; it’ll take on the huge deals and spend the money to make them happen. But it's becoming choosier about which deals it goes all-out for – it's looking for opportunities that showcase new technologies, set the bar high, but don’t break the bank.
So if Oracle isn’t going after the high-end, what is it going after? At this point, it just isn’t clear. The company hasn’t said much of anything, and there isn’t much buzz in the community either, so we’re down to speculation.
I could be wrong, but I just don’t see Oracle abandoning HPC entirely. I think it may call it by some other name or describe it differently, but it will be in the high throughput computing business for the foreseeable future.
There are some interesting angles for it to pursue. Many of its best commercial customers have sizeable HPC or HPC-like workloads that Oracle can now (with the addition of Sun) compete for. I don’t see it passing up those opportunities.
Oracle can also look to specialize on certain subsets of the market and provide more of a solution rather than piece parts. I wouldn’t be surprised to hear of it offering an Exadata-like system that is optimized for, say, seismic or financial services. In fact, Exadata as it stands today is a decent fit for financial service analytic workloads.
HPC can be a profitable business and, in a lot of organizations, it’s growing faster than traditional business processing. From Oracle’s perspective, what’s not to like? ®