Oracle dumping HPC: Genius or foolhardy?

What are the long-term costs?

Blog Quietly - very quietly – Oracle has been dropping out of the HPC market. We’re finally seeing some outward reaction to the company's internal moves with news stories (The Reg here and HPCwire here) discussing Oracle’s retreat from HPC. No comments from Oracle either confirm or deny the move, of course.

I’ve been hearing rumblings along these lines for months; Oracle's absence from SC10 and the scads of HPC sales, engineering and support people getting pink slips all but confirms the HPC disengagement to me. When pressed for an explanation, our contacts started talking about how the company is focused on mission-critical enterprise and, well, maintaining that focus.

Oracle doesn’t seem to understand that HPC is the birthplace of IT innovation. Many of the technologies used in enterprise computing today got their start in HPC, including clustering for scale, the use of Linux for computationally complex tasks, and high-speed storage and networking gear.

Sun hasn’t been a huge factor in the HPC market since the bursting of the dot com bubble and the rise of commodity hardware. But the firm has brought out some innovative hardware along with some design approaches that separated it from the pack, at least to a degree. I’m not sure exactly what will happen to that gear over time; Oracle isn’t going to spend any money developing it, but probably won’t sell the rights to it either. Could it be planning a big ‘Going-Out-of-HPC’ sale? Oracle could pitch a tent next to Lake Larry and sell its stuff on a cash-and-carry basis.

I’m not a big fan of this move by Oracle. To me, it’s exactly the wrong time to back away from HPC. I believe that we’re entering an era where the crossover between HPC and enterprise computing will become even more apparent, with enterprises large and small embracing predictive analytics, visualization, and huge number-crunching in all forms.

Having a vibrant HPC organization would allow Oracle to take a leading role in this trend. Without it, Oracle is just 'the guys with the database stuff'. Supplying customers with Exadata boxes isn’t enough. Oracle also needs to have the expertise to architect and deploy truly massive and complex implementations that can crunch immense amounts of data and supply solutions in real time. This isn’t just data warehousing on steroids; it’s a highly complicated process, and it’s much more difficult to pull off.

Both IBM and HP are using their HPC assets (mainly people and expertise) to shape their offerings in advanced enterprise computing, and they’re deploying their researchers into customer situations. Dell is also beefing up its HPC tool and skill set. By dumping its HPC org, Oracle is putting itself at a competitive disadvantage versus its biggest competitors. Of course, the reduction in headcount and hardware products will certainly save some money in the short term, but it’s likely to be expensive in the long run. ®