I spent several days this week in San Francisco, attending CloudCamp on Tuesday night and the cloud computing conference Stucture08 on Wednesday. Both events were great. CloudCamp was a Barcamp-style Open Space conference with 30+ ad-hoc discussion and presentation sessions on topics all over the cloud computing map. The first session I checked out was a presentation by a Microsoft tech-evangelist on their cloud-database service that I talked about previously. Highly scalable databases for Internet-scale applications is a very hot area of development right now, with Amazon, Microsoft, Google and a host of startups all trying to solve this tricky problem. Although no details were provided, it sounds like Microsoft is gearing up to launch more Web Services in the future – I’d expect to see announcements around PDC in October. The second session was really a group-feedback excerise on Amazon Web Services with Jeff Barr, their lead evangelist. Needless to say I had plenty of feedback to offer . The last session of the day for me was a group discussion on the challenges and approaches to creating an online storage API that could work across multiple clouds. The session was organized by Nirvanix, a cloud storage startup that is working on an open-source PHP cloud storage API. While it would be nice to see a cross-cloud API emerge at some point, I think the market is too new and moving too fast to accomodate any single standard at this point. As it matures I think we will see one emerge, although it may end up being a de-facto standard as other vendors simply adopt the APIs of the market leader(s).

The Structure08 conference the next day was a more typical conference format with a number of presentations, interviews, and panels. For a first-year event I thought it was very well organized and ran smoothly – they even managed to get back on schedule after slipping early on. The quality of the individual sessions was mixed.  Some, such as Werner’s keynote and the cloud-provider panel were interesting, while others like the cloud database panel came off as pure product pitches.

I also found it quite interesting to see how Mendel Rosenblum, Chief Scientist of VMWare, and Greg Papadopoulos, CTO of Sun see their companies positioned in the cloud computing world. Both companies are in a similar position – they sell premium hardware/software for large-scale computing to enterprise, but both are being bypassed by cloud providers like Amazon and Google who are building their infrastructure on commodity hardware and open-source software like Xen for both cost and control reasons. The risk to Sun and VMWare is that enterprises may start moving their applications from their own datacenters running VMWare/Sun and into the commodity cloud infrastructure.

Mendel’s take was that cloud computing today is basically ASP2.0 – and companies that got burned before won’t fall for it again. Mendel sees their VMWare platform as a way for companies to create “mini-clouds” that they can host themselves, gaining the benefits of cloud computing without losing control.

Greg’s take is slightly different – he thinks that just because computing is becoming a commodity, it doesn’t mean you need to run on commodity hardware. He illustrated this with a few examples, like a portable generator versus a powerplant steam turbine. No one expects the electric utilities to run off hundreds of portable generators – they use larger, efficient, and higher quality systems. Of course the problem I see with this analogy is that the performance difference between a portable generator and a steam turbine is huge – many orders of magnitude. By comparison, the performance difference of a white-box Intel or AMD machine isn’t going to differ much from equivalent Sun hardware. There are certainly other factors – such as reliability and maintenance, but at the end of the day, Sun will have to provide significant leadership in $/cycle and cycles/watt in order to displace commodity hardware in the cloud.

Of course for most Jungle Disk users this doesn’t really matter much – in fact, the whole point of the “cloud” is that users (you) don’t need to know or care how it’s built or what hardware and software it’s running. In the bigger picture however this movement is going to change the whole makeup of the computer industry. The wide-scale migration of data and applications into the cloud is going to create new opportunities for some and many challenges for traditional software, hardware, and hosting providers. Online backup in particular is going to rapidly supplant traditional tape backup and offsite tape rotation services within the next few years.