I'll add to Stan's summary: the effort to add DataType API to HBase [0] is driven by precisely this kind of use-case. When applications all agree on the encoding format, this level of interop becomes far simpler and more predictable. As Stan says, we're not there yet, but I'm optimistic ;)
[0]: https://issues.apache.org/jira/browse/HBASE-8089 On Fri, Feb 28, 2014 at 11:13 AM, Stan Campbell <[email protected]>wrote: > Hi Alexey.. You might find this useful: > http://hortonworks.com/blog/hbase-via-hive-part-1/ > > Short answer: Possible but not necessarily performant from the point of > view of Hive. However, if you break out part of your data from HBase > (based on its query profile) you can get excellent performance at a cost of > some disk. > > > On Fri, Feb 28, 2014 at 11:06 AM, Alexey Kharlamov <[email protected]>wrote: > >> Hello! >> >> I am considering deployment of Phoenix for you data warehouse project. It >> seems to fit quite well. But we would like to use Hive to query archive. >> >> Will it be possible to access data in HBase from both Hive and Phoenix? >> >> I read that HBase 0.96 support is on roadmap. But what is timeline for >> release of the new version? >> >> Thank you >> > >
