Abstract: Starting with the building blocks of general file systems (inodes, blocks, journaling) the talk will build up to the HDFS architecture and ultimately the cutting edge HDFS Federation. Some topics covered along the way are: NameNode/DataNode fundamentals, Image/Journal, Namespace ID/Storage ID/Block ID, Secondary NameNode, Rack Awareness, Replication, Read/Write Pipelines, Heartbeats/Communication, Balancer, Snapshots, Decommissioning, Block Scanner, etc.
Speaker: Sameer Farooqui is a Systems Architect at Hortonworks who specializes in designing custom use cases for Hadoop. Previously Mr. Farooqui was responsible for creating proof of concept prototypes using Cassandra, NoSQL and Amazon Cloud Computing at one of the big 5 consulting firms. Sameer also has five years of experience at Symantec where he deployed Veritas Clustering and Storage Foundation solutions to Fortune 500 and government clients.