Yesterday evening marked the last NoSQLSummer we hosted at our offices, and it was decided we would jump from theory to practice by showcasing Lily, our NoSQL content repository built on top of HBase and SOLR. The setting was familiar already: pizza and beer to feed our post-work bodies, followed by intellectual stuff to compensate for that. We were really happy to greet triple the size of the usual audience, some die-hard NoSQLSummer tourists but also a nice group of newbies, or rather: people eager to learn what we have been up to in the past 1.5 year.
I quickly introduce everybody to the bigger picture, and even forget to point them at a first sign of this picture appearing - a change in URL that we feel is pretty important.
In my previous encounters with Lily enthusiasts, I often had to explain in quite some detail what we meant with 'CMS', i.e. that we take the literal meaning, as in Lily being a repository capable of managing content. However, sometimes the subsequent question was: so how does it looks like, where the response is: it's a black-box building block and we're not focusing on providing a typical UI interface. We feel this should be the task of ISVs layering their product on top of Lily, or integrators building one-off scalable content apps on Lily foundations. The reason for this deserves a blog post on its own, but we feel it is hard to do many things well, and a content application typically is a mixture of applicative stuff and content stuff, and oftentimes CMS frameworks claim to do both well while in contrast they do a mediocre job in both areas. So we decided to excel in one area, and leave the other one as an opportunity for partners.
From now on, we'll refer to the bigger picture as 'The Lily Project', and the repository we're currently working on as 'The Lily Content Repository'. For the moment, the repository is the only Lily project we're working on, but that's bound to change next year and the picture above hints at some areas we're currently looking into. Also, we don't think we need to be the ones doing all this, we're sure other parties will be interested in joining the combined Lily effort.
After my shortish intro, Evert explained Lily's concepts, the design of the repository model, the architecture, the design of the RowLog (and why it was needed), as an introduction to a lengthy and quite in-depth demo of Lily's current state by Bruno. The demo was about ingesting a couple of thousands email list archive files from Apache, and make them searchable across SOLR. I'll beg Bruno in redoing the demo as a screencast in the next few weeks, as it shows how easy it is to get started with Lily, and it hints at the hard stuff we're solving behind the scenes (like schema management and index maintenance).
Good times we're had by all, and after a lengthy Q&A we finished long past the scheduled time.
Overall, NoSQLSummer was a great experience for us - finding out about folks working in the same domain we are in, talking about the hard issues and challenging long-standing thought patterns. An experience we should repeat, next Summer perhaps!
Thanks to all who attended, and hopefully see you at Devoxx in November!