Riak: Robust and featureful distributed Key-Value store
Submitted by Prashant Kumar on Thursday, 28 June 2012
Big Data Infrastructure & Processing
In this talk I'll try to bring out unique features for Riak compared to other NoSQL databases. In particular its rich API, high availability and its use as a core building block for big data.
Riak is one of the most robust and featureful distributed K-V store based on the seminal Dynamo paper. In this talk, we’ll briefly cover the underlying concepts of Eventual Consistency and CAP theorem, then moving on to following Riak specific topics and use cases :
- Clustering in Riak will discuss how being implemented in Erlang and being masterless helps increase fault tolerance.
- Quick summary of conflict resolution with vector clock and the data recovery
- Secondary Index and its use as tags.
- Link-walking in Riak and its use cases
- Briefly cover Map-Reduce and Riak-search.
Speaker Bio : Prashant passed out from IIT Kanpur and then joined Yahoo! data team where he worked for about a year. Among other things he worked on the analytics for the new version of Yahoo! front page then.
Currently Prashant runs a leading Big data and Cloud computing startup PromptCloud (http://promptcloud.com) where flagship product is large scale data crawl and extraction and hosted indexing of the data. PromptCloud also deals with Big data analytics but at a limited scale as of now. This talk will be based on many of the learnings at PromptCloud .
PS : I have submitted another talk for ‘Building the infrastructure to handle Big data’ , depending upon which (if any :) ) has enough interest , we’ll pick one.
Link for the other proposal : http://funnel.hasgeek.com/5el/393-building-the-infrastructure-to-handle-big-data