I'll question you on one point: "Use it when you need random, realtime read/write access to your Big Data."
Does HBase now do a good job of random access? I was always under the impression that it did random access adequately, but it's real strength was with scans (based on ordering of keys).
I did benchmarks on a previous version, and the results were pretty miserable (600ms/lookup on a table with several million columns). It certainly sounds like they've improved.
Comments
I'll question you on one point: "Use it when you need random, realtime read/write access to your Big Data."
Does HBase now do a good job of random access? I was always under the impression that it did random access adequately, but it's real strength was with scans (based on ordering of keys).
Here's an informative presentation about how HBase has (massively) sped up random access, as of version 0.20: http://www.docstoc.com/docs/7493304/HBase-Goes-Realtime
I did benchmarks on a previous version, and the results were pretty miserable (600ms/lookup on a table with several million columns). It certainly sounds like they've improved.
I stand (very happily) corrected.