[irq]: techie interrupted08/10/2012
“ In the mid of 2011, when the Diablo release was under heavy development, we decided to use OpenStack as our underlying system of Sina IaaS public Cloud, Sina Web Services(SWS), and another strategic product besides Sina App Engine(SAE), which is developed by my former team members and already the most popular public PaaS cloud in China. But then OpenStack was full of bugs, not very stable and not ready for production deployment, and also lacked some essential components, such as billing, monitoring and load balancer etc. So we invested several engineers to do bugfix, to implement new features and to design necessary services. In the beginning, we forked an internal branch from a particular commit of OpenStack, and had much development on the internal branch. Later we found that it is a little difficult to merge upstream updates to our own branch, if this condition did continue, our project would be dangerous since it would go more and more far way from the official projects, and we would finally lose the community and the ecosystem. So we stopped the trend immediately and cut down our own fork. Instead, we joined the community, collaborated with gurus around the world, and combined the requirement of our own public cloud projects and need of OpenStack community, so that we could be able to avoid duplicated development, and it has become a win-win game for the community and my employer. In fact, we have contributed all our bugfixes and feature improvements to upstream. We also opened sourced our own implementation of biling(Dough) and monitoring(Kanyun). We benefited a great deal from this change. „
Quote posted at 22:54 Comments
blog comments powered by Disqus