Few days ago I flew to London to give a talk about Apache Spark at local Scala User Group. This was both enriching and fun experience. Not only I could catch up with some old friends, but I was also able to meet new ones. I also had time to do a lot of sight-seeing, thanks to my lovely wife, who guided me through all the cool places that London has to offer. We’ve also had couple of unexpected surprises, one of which was actually showing up late at Skill Matter’s venue due to the the fact we’ve completely messed up our time table. Eventually we were forced to take a taxi, which, as one can imagine, is extremely expensive in London.
Around 120 people showed up at the venue. I was happy to learn that the audience was pretty much familiar with Map Reduce concepts, thus, even though I’ve stared with my original talk “Apache Spark 101”, I was able to quickly skip through a lot of slides and focus on Spark’s core and internals.
I’ve finished my talk 45 min faster then expected. However seeing that the guys from the audience seemed to rather enjoy it, I’ve decided to improvise. I’m currently working on two seperate presentations regarding Apache Spark: “Apache Spark – a bit about streaming” and “Apache Spark – when things go wrong”. Both however are in status ‘work in progress’, far way before we could considered them finished. Nevertheless I’ve decided to go with what I had. For the talk about Spark Streaming at least I had some slides prepared and some running code examples. The talk about Spark internals however is merely planned in my head and I was forced to fully improvise, creating examples on the fly, moving my hands in the air, trying to visualize the concepts I was explaining. This was new experince but I hope I’ve managed.
Overall I think all went good. I’m still waiting for the feedback from the audience. After the talk I was approached by the leader of Apache Spark User Group that runs in London and he said that they would be intressted hearing more about internalls of Apache Spark. So it seems I might be revisiting London sooner then expected 🙂
As always, the ‘after party’ was knowledgeable. Having few beers with the guys from the commmunity can sometimes bring more news than a 3-days conference. I’ve learned for example that Netflix has actually not first but second larges Cassandra cluster deployed in production. The first place goes to the company whose name here will not be mentioned, but let’s just say, that you won’t be able to have a bigger Cassandra cluster then those guys – period.
So to sum things up:
1. London is awesome, it has a lot to offer. Something that I’ve only experienced once in NY
2. People living there are joyful, approachable and very friendly.
2. Scala meetup brings developers with a lot of experience in creating distributed systems.
3. I was able to provide presentation about Spark’s internals, even though I was not fully prepared for that