Hi HBase-dev, I'm working on improving caching in HBase as a part of academic research. I was wondering if you could help me answer some "how HBase is used" questions.
1.Are there instances of real-time applications that rely on HBase's datanode caching directly? I realize that many real-time applications employ a caching layer (like memcached) in between their stores and front-ends. Are there cases where HBase is used directly without an intermediate caching layer? 2. If so, why is an intermediate caching layer not employed? Note: by "real-time", I mean workloads that require immediate responses to a user: say a web application or real-time analytics (unlike MapReduce jobs that may be long running over several hours). Any advice would be greatly appreciated! Thanks, -- Bharath Ravi
