Hibari Presentation to Chicago Erlang User Group: Wednesday 25 August, 2010

I’ll be giving a presentation about Hibari to the Chicago Erlang User Group (and hopefully some ChicagoDB folks) on Wednesday, 25 August, 2010 at 6pm.  Orbitz will be hosting the event, please see the CEUG event at MeetUp for more time and location details.  It looks like I’ll be arriving in O’Hare shortly after noon on Wednesday and will be returning Thursday morning.

Here’s a copy of the abstract that I submitted for the talk:

Hibari is a distributed, fault tolerant, highly available key-value store written in Erlang. Data replicas are managed by chain replication rather than the more common quorum voting technique used by Riak, Cassandra, and others. Hibari supports micro-transactions (atomic updates to multiple keys in limited circumstances) and cluster resizing (both growing and shrinking the cluster). Hibari clients can be written in Erlang as well as C, C++, Java, Python via the UBF protocol and soon in other languages via Protocol Buffers and Thrift.

Hibari was released to the open source community under an Apache License v.2.0 and is available at http://hibari.sourceforge.net/. Hibari’s largest commercial deployment is a Webmail system in Asia that is managing approximately one billion email messages.

My talk in Chicago will be the Hibari’s first public presentation. I’ll talk about its major features, how chain replication works, and how Hibari’s sponsor (Gemini Mobile Technologies) will manage code patches from the open source world. I’ll leave the rest to audience questions, with just-in-case material for demo’ing things like cluster expansion/contraction and developing a small client application that uses micro-transactions.

Many thanks to Martin Logan, Chris Duesing, Eric Merritt, and Jordan Wilberding for the invitation and for all of the cat herding heroics that an active user group needs.