Mårten Mickos – Open Source Cloud Platforms
Mårten Mickos of Eucalyptus opened day two with a presentation on open source cloud platforms. He opened by naming the big four open source cloud platforms – OpenNebula, OpenStack, CloudStack and Eucalyptus - the ‘four beautiful open source sisters’. He drew further analogy by explaining that Starbucks is the public cloud – you know exactly what you can get, and that the home espresso machine you have on your bench is the private cloud. They serve different purposes.
He went on to explain the different types of cloud – public, private, hybrid, and mobile – and asserted that everything will go mobile eventually – a prediction I wholeheartedly agree with. As a Eucalyptus rep, he was obviously outlining where Eucalyptus’ services sit, and he positioned them between the public cloud and data centre – where people might be making their first move from the data centre to the cloud, or in reverse where there are some services they wish to bring back in house.
Mickos’ stated that Eucalyptus understood the desire of developers to have their workloads liberated from the underpinning cloud platform – and to not be locked in to any one platform. This ability is scaring some hardware vendors, as increased core and node utilisation from leveraging cloud services means that less hardware is required.
Mickos went on to opine the four open cloud offerings were a source of better innovation through cross pollination – their competitive collaboration made each strive to be better. In conclusion, Mickos stated that the open cloud provides developers with freedom of environment, scale and deployment.
Monty Taylor – Growing an open source community
Taylor, of OpenStack fame, demonstrated how they grew their community through a fundamental commitment to openness. This starts from the basic belief that anyone can participate and contribute in the project, and that there is a sense of liberte, egalite and fraternite throughout. The freedom of the project also needs to be assured through ascribing an appropriate license, and the governance models and process of the project also need to be egalitarian. While a benevolent dictator (in the form of an individual or a large corporate sponsor) may seem like a positive thing, this can be offputting for meritorious contributors. Taylor explained how open, participative design summits – copied from the Ubuntu model – were also used to help cohesion within the community. He further explained that the repositories for a project also need to be transparent and open, and that code reviews were necessary to ensure quality, but they also needed to be transparent and open in nature.
The key takeaway here was that in order to grow a community in opensource, everything about it needs to be open – the code, the governance, the processes, the contributions etc.
Matt Asay – Picking the next black swan in open source
Matt Asay, who works with 10gen (a mongoDB company), presented on how to pick the next ‘winners’ of open source. He opened by sharing some of the biggest ‘black swan’ events of the past 40 years – such as ‘no one will every need more than 640kb of RAM’ – and highlighting that it’s difficult to predict the future.
One key principle that’s often employed is to ‘follow the money’, but Asay demonstrated that this is not always true. In open source, in contrast with the commercial sector, money and profits are not always a harbinger of success. Some open source companies make massive technical breakthroughs without being financial successes.
Instead, he encouraged those interested to observe the user and developer ratios of projects – and quoted a figure that for every 1000 users, there are 10 bug reporters – and one developer. So the ratios are important.
Other factors to consider when picking open source winners included whether the project was prepared for participation by having assets such as documentation, modularity, accessibility to code and to knowledge of the project, a solid codebase and a license to fit the need. Where the code was hosted was another factor, and Asay highlighted how the market dominance of SourceForge and SVN was being usurped in recent times by GitHub.
He also encouraged us to follow the developers and the data – as big data is getting bigger all the time, with technologies such as Hadoop, NoSQL and analytics having a larger role to play int he future. Big data is big on processing, and big on storage, and this has forced companies such as Google, Facebook (viz. Cassandra) and Amazon to write their own stuff – big data was the driver. From there it was not a massive leap to show how to pick open source winners by following job market trends. Buzzwords such as HTML5, iOS, Android, Puppet, Hadoop et all weren’t even invented 2-3 years ago.
He concluded by stating that in open source, we build what matters. We innovate in technology – and open source is focussed on big issues – not – for example’s sake – on finding ways to get people to click more advertisements.
Imad Sousou – Linux at Intel
Sousou opened with an explanation of Moore’s Law, and how it has more or less held true for the last forty years. He then applied the law to the automotive sector – and if a VW Beetle followed Moore’s Law, it would go at over 300,000 km/h and run for 5cents a week! The type of constant innovation that allows to evolve at such a rapid pace takes a lot of dedication, commitment – and investment.
Sousou demonstrated how Intel had played a key role in Linux communication and many open source projects such as Wayland, dLeyna and many others, with the overriding theme being the development of apps in web technologies. Here, two challenges are still present – API completeness and performance. The W3C APIs simply don’t cover what an application needs to do, so Intel have helped create the System Application working gorup within the W3C to improve this. They are also working on web technologies performance, to help improve things like fluid animation. Their investment in automotive Linux is also to be noted – and one wonders whether it will be long before we have Linux not just in the the desktop, the data centre and in our mobile devices, but also under the bonnet.
Ralf Flaxa – Enterprise Linux Evolution
Flaza, VP Engineering at SuSE opened by stating that he didn’t want to give us the standard sales pitch. Instead, he told us a story of how he became involved in Linux – and it was all because he wanted a serial driver. From his first Linux Kongress in 1994, he still feels the sense of collaboration and community, even though many members of the Linux ‘family’ are now his direct competitors.
Much of his talk echoes previously covered themes – such as what it means to be ‘open’. To be truly open means being open in many ways – open source, open licence, open community, open governance, open repositories, open to invite the competition and open to contribution.
He gave a number of hints on how to achieve this such as
- grant influence only to contributors
- welcome contribution in any form – documentation, code, money, testing
- encourage beginners and lower the barrier to entry
- remember to give credit and recognition
- strive for the best possible code base
- keep things simple by modularising them and breaking them down into digestible bits