Last week I traveled to Cologne, Germany, to attend the NoSQL Matters 2013 conference. There were lots of great talks, including both beginner and intermediate talks about all things NoSQL. It’s also really good to see the audiences have more familiarity with a variety of database technologies, and are more keenly aware that they have to choose wisely from the many different kinds of NoSQL databases out there.
That’s a point we try to make here, too. Sure, there are lots of scenarios where eventually consistent key-value and document databases are a great solution, just like there are still lots of scenarios where relational database technology is still appropriate. But eventually consistent databases can require your application to manage conflicts, and this may or may not be easy. Sometimes, it’s just easier and more natural to write code that relies upon ACID transactions to guarantee strong consistency.
I gave a talk about using ModeShape as a consistent, in-memory, hierarchical database. Again, it was primarily an introductory talk that covered most of the features that distinguishes it from other database technologies, including relational, key-value, document, and even graph databases.
I also attended several really interesting talks. One of them was “Processing a trillion cells per mouse click” by Alex Hall. Alex talked about how Google needs to very quickly process vast amounts of data so that internal tools can enable Google’s employees to interactively drill down into ad click data to look for various usage patterns. The resulting solution uses a variety of interesting techniques to optimize the indexes to support these drill-down queries. You can read more about it in this paper.
Another great talk was given by Simon Willnauer, who showed how easy it is to use ElasticSearch to query and search your data, even when that data might be stored in another system. (We’re considering having ModeShape use ElasticSearch for our indexing and search uses, so this was quite timely.)
Tug Grall gave one of my other favorite talks. He’s from Cloudbase, and this talk focused on how you use Map-Reduce style JavaScript functions to build indexes from your documents. I’ve talked with Tug before, and it was great to see him again.
Kudos to the NoSQL Matters 2013 organizers for doing such a great job organizing and running the conference . The talks were quite good, it was well attended, and the meals and snacks were above par. I also have to say that Cologne is a fantastic city with incredible charm and history. What a great place for a conference!
Filed under: appearances, features, jcr, presentation