CloudCamp

  by Andy Ormsby | 13 Mar 2009 | Permalink |

I went to CloudCamp last night. For those who don’t know, CloudCamp is a series of UnConferences about Cloud Computing (as opposed to a conference about camping in places with inclement weather). You can find more information here.

This CloudCamp was the third to take place in London. The previous London CloudCamp was my first and was held in the crypt of St James Church, just off Clerkenwell Green. By comparison, last night’s event at the Queen Elizabeth Conference Centre, was huge, though the number of people talking about real experiences doing stuff with Cloud Computing was a little limited.

As with the previous CloudCamp, this one started with a series of “lightning talks,” 5 minute spots, strictly enforced, where the speakers try to say something interesting, informative or just provocative. One of the rules is that the talks should not be product pitches, something that clearly causes some speakers to struggle. This is good in one sense, because the product pitches are usually dull and feeble, but bad because it would be good to learn a little more about some of the products people are building.

Two highlights:

Mark Masterson from CSC gave an excellent talk (and was also one of the moderators for the Enterprise Clouds stream that followed the lightning talks). Mark talked about security, essentially claiming that it probably isn’t much different in the cloud than in your own data centre. I’ve revised my opinion of CSC. I had categorised them as yet another giant consulting/outsourcing firm, but there’s clearly something more going on. Mark has a blog which looks worth following.

Joe Baguley of Quest Software finished the lightning talks with a provocative take on the whole cloud movement. The gist of his message was that people don’t know enough about what their applications really cost them to justify moving them into the cloud. This rings true from my experience. All too often, the people I talk to don’t have an accurate view of what servers they have in their data centres, let alone a decent mechanism for calculating the cost of the applications that run across those servers. So if you’re going to use cost saving as a justification to begin migrating stuff into the cloud, you’re probably stuck.

It is a familar story. If you are embarking on any transformation exercise, understanding what you have before you begin is a really good idea. Increasingly, that is what we find ourselves doing at Tideway: helping companies get a grip on their data centre infrastructure so that they can start reshaping it.

The session on Enterprise Clouds, one of four streams which followed the lightning talks, was slow to start but covered lots of useful ground.

One of the concerns raised is that of being locked in to a proprietary platform. If you’re software supplier goes bust you have a problem, but the likelihood is that things will keep running and your data remains safe in your data centre. But if the data is in a proprietary cloud and the service provider goes out of business, perhaps you will go out of business too. The fate of customers of Coghead, who have until the end of April to get off the platform, is a warning to anyone prepared to bet their business on someone else’s. Perhaps that’s why, for now, some of the most talked about success stories in cloud computing are for problems where the commitment can be short term like HPC, as demonstrated by Eli Lilly.

An interesting point is that while lock-in is more obvious for fabric-based clouds like Google AppEngine or Force.com, it may also exist for less obviously proprietary instance based cloud platforms like Amazon, possibly in insidious ways. Mark Masterson gave the example of a company that has accumulated enough data in Amazon’s S3 to make the simple volume of data a practical barrier to migration elsewhere.