MongoNYC 2013 post conference report

By June 27, 2013 August 18th, 2022 No Comments
ObjectRocket skyline

10gen has been doing MongoDB events for years now, and I have attended quite a few of them. The events have always been great. However, I have to say this years MongoNYC was particularly amazing.

I left New York thinking the caliber and level of the discussions, presentations, contributors, and partners has really matured and this event was clearly the best yet.

Of primary excitement for the ObjectRocket team was the announcement of the Partnership with 10gen. At ObjectRocket we are very excited to collaborate with 10gen at new levels to bring an even more fantastic MongoDB experience to the cloud. Personally for me this is a formalization of the collaboration we have had in the background for years. It’s about time we jointly brought a true Enterprise quality high performance cloud offering to the masses. I am super proud to be part of it and look forward to an amazing future of helping customers maximize MongoDB in the cloud.

The folks over at Silicon Angle also grabbed us for a brief interview. Chris Lalonde and myself talk ObjectRocket, the 10gen partnership, Database as a Service, and how ObjectRocket got started.

I presented “Scaling MongoDB; Sharding Into and Beyond the Multi-Terabyte Range”. It was very exciting to have a great turnout, but more importantly, I was very impressed the caliber of questions from the audience. This year the audience seemed to really be thinking at a new, and deeper level about MongoDB deployments as well as having some good actual experience in many cases. It does appear that folks are beyond largely beyond adoption/evaluation phases and being challenged with the day to day development and devOps duties of MongoDB applications. My slides are available in Github and on the 10gen presentations page. I will also post the video up when it’s done being edited. It’s worth noting at this point that Richard Kreuter mentioned I omitted Secondary Throttle in my talk. He’s right, slides will be updated. Thanks, Richard!

The one topic that kept coming up in my discussions were shard keys. How do choose them, how to alter them once they are set as workloads may change, choosing hashed vs range, and more. In my talks I focus on the concept that choosing shard keys is generally a compromise, and many folks where struggling in how to compromise intelligently. This is an area that I think ObjectRocket can help with, and we plan some tools that should aid in making good choices.

Other top issues seemed to be:

  • Hashed based sharding. What is it, how does it work, how to effectively use it, and overall joy that it’s available now in MongoDB 2.4.x
  • Compaction/Fragmentation. Lots of discussions both in the Masters group on Thursday as well as with end-users and customers. ObjectRocket currently supports online compaction on our service using Replica Set members. Eliot mentioned powerof2Sizes which, to some degree, has been effective for our customers. I think powerof2Sizes needs more general visibility because folks don’t know how it works and it’s effect on fragmentation in general.
  • Working Set. Lots of discussion around I/O, sizing memory, estimating working set, and details around using solid state disk (SSD) to aid in performance. Clearly this problemset is something that ObjectRocket alleviates by it’s very architecture. We listened to folks discuss the challenges of dealing with I/O in MongoDB. Great discussions overall. It seems like a blog post on the workingSet estimator may be in order.

We had great traffic at our sponsor booth. We first became a sponsor for MongoSF and now MongoNYC 2013. This is a really great advantage of being part of Rackspace, we can participate in the community at a new, higher level. The amazing Racker Kathleen came out to help us be able to speak to potential and existing customers alike. Plus, it was her to thank for the awesome Lego® R2D2 and Raspberry Pi give aways.

We took advantage of the time at the conference to meet with a couple of key customers, gathering feedback on our service as well as discussing a new addition to our service that we are just starting to work on now. We are gathering feedback and getting customer perspectives on this new offering as we start to lay down the initial design. More to come, but for now we have great info, and are taking this back to our offices in Austin where we can get some coding done.

Lastly, what conference wouldn’t be complete without an after party. All I can say is wow, we had some fun, some beer, and had a great time. Nicely done on the entire conference to the 10gen team. Best one yet!