Today’s interview features Steve Loughran, chair of the Hadoop Futures track. Steve is a member of the Hortonworks technical staff, working for CTO Eric Baldeschwieler. In this role, Steve explores leading edge options for the Hadoop platform.
Tell us about the Hadoop Futures track.
We’re looking at the future of Hadoop – what is coming next in Hadoop, HBase and other parts of the system.
Who else was involved in the selection process for your track?
Johannes Kirschnick, who has recently started doing a PhD at the Technical University of Berlin after working at HP Labs for a number of years, and Bill de hÓra, heavy user of Hadoop and related technologies in production. Bill’s based in Ireland and I’m in the UK, so we had people from three countries participating in the decisions.
What was the hardest part about the selection process?
Hardest? Being ruthless and dropping my own talk from the list. I was going to look at future hardware trends and their impact on Hadoop (read this if you’re interested), which is something we care about -e.g. what will a low-cost SSDs bring to Hadoop? will servers with many-but-less-powerful CPU cores change job execution, etc. I think it’s profound, but there were other talks that were of more immediate benefit to the audience, including topics focused on where HDFS is going, the future of HBase, Hadoop on Windows, and the like. Those talks are going to be more important to most of the attendees of the conference, which is why I had to turn down my own proposal along with many other good ones.
What takeaways do you expect for folks attending sessions in your track?
Quite simply: where Hadoop is going and what opportunities are coming next.
What are you personally most looking forward to at Hadoop Summit Europe?
Meeting more people in Europe who are using Hadoop. Most of the development of Hadoop takes place in the US, which is where most of the largest users are based, but there are really interesting projects starting up in Europe. I’ve been fortunate to know some of the people working in this area over the last few years but I’d love to meet others, discuss what they are up to, understand what problems need to be addressed in Hadoop, and generally have fun. At the 2012 Apache Conference, I met two people who had to deal with a dead whale falling through a ceiling by their Hadoop cluster, which is the kind of thing you have to hear from the source to believe. The cluster stayed up, incidentally.
If you could attend one session from another track, which session would it be?
Allen Wittenauer’s talk on Operations at LinkedIn.