Keeping Grid and Cloud Computing Seperate

There is nothing I enjoy more then stirring up controversy and lately it seems to find me where ever I go. As some of you know I've been attempting to organize a Wall Street Cloud Interoperability Forum for this March. In my planning, I seem to have stirred up some great debates among the more established high performance computing & Grid computing organizations who have focused on the banking and fiance industry.

In a conversation yesterday with prominent wall street grid advocate, he bluntly said that no bank would use external (cloud) resources anytime in the near future and that a uniform cloud interface was a hopeless cause (I'm paraphrasing). This is pretty much exactly the opposite from what I'm hearing from several high level IT folks within the banking industry. And actually quite the different story from what Craig Lee, the President of the OGF has said to me. According to Lee, the intersection of cloud computing is a critical area for the OGF and one of their main focuses going forward. I think the hardest part is going to be convincing the established HPC/Grid community that cloud computing and grid are not one in the same.

(I should also note we're putting on a joint CloudCamp with the OGF at their OGF25/EGEE User Forum in Sicily this March)

To shed some light on the whether banks are truly embracing cloud computing I had recently the opportunity of speaking to many IT managers on Wall Street over the last few months. In one such conversation early this week with a major German bank, I was told that most banks are being forced to do a lot more with a lot less money. They see the ability to outsource the so called low hanging fruit such as test/dev, performance testing, continuity, as well as various customer facing services as immediate opportunities that will allow them to better utilize existing infrastructure while moving less important, yet capacity intensive applications to the cloud. Also the the concept of a unified cloud interface (UCI) that enables a single (standardized) programmatic interface to both internal and external resource is of particular interest. I've received no less then a half dozen calls on the subject of UCI in the last two weeks alone. The fact is, I've heard the same story from several of the largest banks on Wall Street, it isn't that cloud computing may happen, but that it is and there is a lack of tools to enable this migration/transformation. The opportunity would seem to be in addressing the intersection of legacy IT infrastructures with the hybrid infrastructure of the future. I'm not alone in this thinking, notably Cisco and Sun have stated similar opportunities. (I should note, both Cisco and sun are involved in my cloud interop activities)

I think what bugs me about the HPC/Grid guys is that they seem to be driven by academic motivations of using distributed computing as a kind of universal problem solver. It's been almost 10 years since Grid technology was first discussed and yet we seem no closer to any kind of wide industry adoption. In a little over two years, cloud computing has been able to do something grid has never been able to accomplish, become the next big thing. Yes, I know hype doesn't make a sustainable industry, but it certainly helps.

The Grid computing community's problem is it doesn't appear to be driven by saving money or any broad or practical business benefits, but instead by solving very particular problem sets. I feel that one of the major benefits of cloud computing is in the fact it does address a much broader opportunity, one that touches upon just about every aspect of a modern IT environment. I also find it interesting that unlike the Grid/HPC community which is predominately driven by academia, cloud computing is being driven by business and more importantly actual business problems. Businesses who for the first time are able to tap into near limitless opportunities driven by cheap and easy access to compute capacity.