Wednesday, April 22, 2009

There and back again ... a personal journey.

Back in 2006, the company I ran, Fotango, publicly launched a JavaScript application platform that ran on an utility computing environment. It was known as Zimki. It was a forerunner to what today we call PaaS (Platform as a Service).

Zimki developed a loyal and forgiving following. Platforms on the internet were new and we were learning. Amazon EC2 also launched (a few weeks later) and we felt we were in good company. However, we knew that creating competitive marketplaces that gave users a choice in providers and easy portability between them was critical for the success of this future industry.

Although this wasn't a new idea - the same effect had happened in networks and many other fields - it was the reasoning behind the decision to announce the open sourcing of Zimki back in October 2006. We planned to create the "infrastructure for a 'national' grid for Zimki environments" with choice for users and competition between providers.

Yes, this meant we were going to give an "easy route to competitors into this environment", but that's the point - to create a wider market.

Our future was set, a marketplace of providers with freedom of code and data based upon standards implemented through open source code. It made logical, economic and business sense, Fotango was a profitable company and there was the potential for huge future revenue streams. Azure, Force.com and Google App Engine were still "visions" and we were a pioneer. Make the right moves and we would become the dominant platform in this space.

Unfortunately, I miscalculated. Our parent company Canon (or more importantly their consultants  / advisors / whoever ) didn't see our "cloud" computing vision as aligned to its core business. We were toast, the open sourcing was stopped and Zimki ended.

Without an alternative provider, our users were unable to move their code and data. This was one of the problems we had tried to solve and alas the same thing has happened again with CogHead. Anyway, I no longer work at Canon. Today, I work at Canonical.

Canonical embraces open source and the "cloud"; it understands that this is a future direction of technology and it sees this change as core to its business. In the next release of Ubuntu we've included Eucalyptus into the distribution. This means that anyone can create their own private cloud which matches AWS (e.g. the Amazon EC2 / S3 API) using open source technology. The reasoning for choosing Eucalyptus was simple: pick the emerging standard and find an open source reference implementation of it.

We're also releasing official Ubuntu images for AWS. We want to help and encourage our users to build in Amazon's cloud along with building and experimenting with their own clouds. We also want to see multiple providers develop around AWS APIs, and portability between them.

Amazon EC2 API (et al) are the emerging de-facto standard that the market has chosen. Hopefully, with Eucalyptus in the Ubuntu distribution we will see the market further adopt the EC2 API in private clouds and, ultimately, actual moves towards an open cloud.

This is my one grumble with the open cloud manifesto. It's good to talk but the ideas of an open cloud environment have already been heavily discussed over the last three years.There's no need for a manifesto that rehashes old ground; what's needed is the adoption and promotion of standards based around open source reference models.

Fortunately, in a bizarre twist, the open cloud manifesto might actually achieve this. The 170 odd companies who have signed up to it have all agreed "to adopt existing standards wherever appropriate", to use "standards organizations" and to address portability through standards.

With further market adoption of AWS supported by the inclusion of Eucalyptus into the Ubuntu distribution, we might see a push towards IETF recognition of a market chosen standard. At which point, if the companies behind the open cloud manifesto are true to their word we should see widespread adoption of the EC2 / S3 APIs as the standard.

For Amazon this is the strategic equivalent of checkmate against any new major entrant, as it provides a mechanism to retain competitive advantage by being at the forefront of any standardisation effort. For the community, it is a step towards a future of portable environments between providers and the promotion of open source as the dominant model in the cloud computing space.

This is what I tried to start with Zimki all those years back. It does create a feeling of deja vu to be back in a similar position. At least this time, there will be no broken promises and it's going to be done right. The technology preview of Ubuntu Enterprise Cloud which combines Eucalyptus into the Ubuntu Server Edition will be released tomorrow (Thursday 23rd April)

--- Update 13 Nov 2013

Today, Eucalyptus continues with its AWS compatibility story and is close to passing 100K cores in one installation. Amazon continues to not only dominate but accelerate away. The open cloud manifesto became a hopeless joke and alas most of the old competitors still fight against the inevitable. 

Fortunately, CloudFoundry plays a solid open source platform game and is full of promise. Canonical has grown and vastly outstrips RHT in the cloud. I was so wrong back in 2008 about my fears of RedHat playing a clever game that I couldn't see - turns out that they were fighting blind and didn't know what was happening. 

Various standards bodies are trying to impose their will on a market that has already decided, Javascript continues to rapidly grow and as for Canon, well ... 

The technology that was lost - Borg, Zimki, ClockThat - have today's equivalents representing many billions of market capital. That's not even including the 3D printing stuff, utility billing engine, the facebook platform and numerous other efforts. In another universe, given a huge amount of luck then Canon might have done quite well in this cloud space or maybe not. As for the embarked upon television products, well I did warn about yield ... gosh, I do dislike big name strategy consultants aka kids with 2x2 diagrams of well intentioned drivel.
Post a Comment