Greetings. I hope you are having a fantastic holiday. This is my last post of 2008 and I wanted to recap the year. This is the first time I’ve attempted this, so hope you’ll bear with me. This is not the typical 2009 predictions. I already gave at the office on that one. Instead, I thought I would chat about what I did this year in relation to cloud computing.
As you may know, I was building a cloud computing management startup for most of 2008 until I took a position at GoGrid in late November. That startup, CloudScale, ultimately was unable to get off the ground, but had a great run.
I had the first inklings of CloudScale in December 2006, when James Gartrell and I built a very early working prototype of a similar technology called the Virtual Server Room (VSR).
The VSR was a way to dynamically create and provision a full back office ‘stack’ for startups: Active Directory, Exchange, Trac+Subversion, and a network-installation server (for Windows + Linux). This prototype worked with VMware exclusively and took 5 minutes to configure an entire stack. What would normally take an expensive consultant 1-2 weeks to develop took us 5 minutes using virtual machines, savvy, and a relatively small amount of code. This was really the ‘big a-ha’ moment for me (as opposed to some of the smaller ones). We ultimately ditched the VSR, but it paved the way for CloudScale and showed how much that infrastructure was being transformed by technologies like virtualization.
EC2 in Da House
I evaluated EC2 about the same time we were building the VSR prototype and as we were jettisoning that product, the idea of consuming on-demand virtual servers was a literal siren call. We could reuse many of the VSR ideas and thinking, but no longer have to run our own infrastructure. In looking around, I reached out to Luke Kanies of Puppet fame, and through conversations with him and others the final nuggets that ultimately created CloudScale were formed. By late March of 2007 CloudScale had been designed and by end of Summer 2007 the very first prototype was done.
We were so early to market that no one was talking about ‘cloud’, the product had no official name, and I knew of no one building automated management systems around Puppet+EC2 at the time. Due to a bunch of setbacks, CloudScale was backburnered from Sep through Dec of 2007, but come January of 2008 we picked up steam again.
Rebuilding Momentum in 2008
As 2008 opened, CloudScale started to build back momentum. Interest in EC2 had picked up significantly, some competitors could start to be seen, and the word ‘cloud’ was becoming common place. It was about this time we officially branded the product, took on our first reference customer, and iterated through a major revision almost every month.
The Cloud Starts to Boil
Right before summer time, the term “cloud computing” was popularized. You can see a significant and ongoing upswing in google searches for the term. To give you another sense of how much it grew this year, just look at the Wikipedia stats where it’s currently ranked at 1134 of *all* articles. In January of 2008 the cloud computing article on Wikipedia saw an average of roughly 800 pageviews per day. By December of 2008 that jumped to somewhere around 4,500 pageviews per day. We were well into the ‘hype cycle’.
Given this climate, CloudScale was starting to look like a ‘sure bet’ to me. Our first reference customer had driven us to support not only Linux on Amazon’s EC2, but Windows on GoGrid as well. At the end of summer ’08, to the best of my knowledge, we were the only multi-cloud, multi-platform cloud management system available anywhere.
Demise of a Great Idea
Unfortunately, towards the end of summer, CloudScale took a series of setbacks, all within a two week period. Funding and founder problems made for a double whammy that required some creativity. With a reduced team and a more solemn perspective we started looking into open sourcing part of the product and started planning for a 6-9 month time horizon before attempting to look for financing again. This seemed like an actionable plan and we were trying not to die as a startup.
About this time the market did it’s nosedive and the 6-9 months started to look more like 12-18 months realistically. Also, it was unclear how successful we would be at continuing to bootstrap on consulting revenue from neoTactics given that consulting dollars seem to be the first to go for many organizations trying to save money in a downturn.
While I think an economic downturn is a great time for a funded startup to create opportunity and thrive, it’s a terrible time for an unfunded and bootstrapped startup to try and survive.
As this was becoming clearer to me, a variety of opportunities arose. The most interesting of which was joining the GoGrid as the technology visionary on the executive team.
Onwards and Upwards
Tabling CloudScale for the foreseeable future I made the transition to GoGrid. GoGrid launched their public BETA in early 2008, was one of the more credible EC2 contenders, and I had some great experiences with the team and the CEO, John Keagy, in particular.
I had a strong desire to continue to be deeply involved in cloud computing and a ton of expertise from having been heads down developing for both EC2 and GoGrid for well over a year. GoGrid needed someone who was a technology visionary with cloud expertise so this was a matchup that made alot of sense.
It’s only been a month, but so far GoGrid has really been a great experience and I’m sure it will continue to be so in 2009. This is a team and an organization that is deeply committed to the cloud computing vision. It’s exciting to be working with folks who ‘get it’ and are running as hard as they can while also improving the organization itself as fast as possible. This is a growing business that knows how to grow. Re-tuning the engine and changing the tires while driving your car is non-trivial to say the least.
So, 2008 was an interesting year for me. I saw cloud computing early and positioned myself as best possible. Ultimately, it paid off, but not in the way I had planned. While CloudScale is down for the count, GoGrid is just taking off and the possibilities are staggering. 2009 promises to be a big year for GoGrid, cloud computing, and infrastructure. I’m very much looking forward to it.