Andrew McLeod
on 19 October 2015
Earlier this month, myself and Kevin — another member of the Juju big data team — attended the 2015 Strata+Hadoop World conference in New York City. As my first big data conference since joining the Juju team, it was an impressive experience, both running the competition pod at our Canonical booth and hearing speakers go over the technical and not-so-technical details of their big data implementations.
While many of the talks were not particularly technical from an operational perspective, the Q&A time at the end of each talk provided — as you would expect — opportunity to dig more deeply. The Walmart talk was particularly interesting with respect to how their culture has changed to allow employees to more easily access and utilize their big data platform to allow quick and easy deployment of a cluster which a user could then populate with whatever data their particular endpoints would provide to them. This allowed people to take their innovative ideas and test them almost immediately on real data without having to wade through paperwork and wait for approvals. I couldn’t help but think how useful Juju would be in this environment — not necessarily replacing something like Ambari or Cloudera Manager, but supplementing it to allow data scientists and other end users to create test/production clusters without taking time from an operations team.
All of the Netflix talks were absolutely packed, showing that big names draw big crowds. From our perspective, it’s very exciting to hear about services that are shaping the future of big data within large organizations like Netflix and Walmart. It gives us insight into projects that are gaining traction in the big data world, and helps us plan future charm development to make sure we have those offerings represented in Juju. Spark Streaming and support of the Parquet data format are two exciting topics for us to work on during our upcoming development cycle. We’ll also be keeping on eye on Kudu and expanding our ingestion endpoints with Kafka and Flume.
Attendance at our booth was consistent throughout the event. We had lots of interest in our live demos and want to extend a big THANK YOU to our partners for helping us showcase some of the solutions made simple with Juju:
Another special thanks to everyone who came over to the Canonical booth to chat with us — it was great meeting you all. Overall, we felt that the conference was a huge success. We look forward to putting our new knowledge to good use and hope you’ll join us in extending the Juju Big Data ecosystem!
