The idea is to be able to run HBase directly on top of WASB, rather than HDFS. The data files are managed and stored in block blobs which are suitable for this workload. The WAL files are stored in page blobs.
For that regards, we have been working on getting the durability / recovery semantics that HBase requires from the underlying filesystem to be implemented on top of azure page blobs. Things like atomic renames, recoverLease, etc for ensuring fencing, etc. I can go into more details if anyone is interested. I believe there is already a preview of this here: http://blogs.technet.com/b/dataplatforminsider/archive/2014/06/06/announcing-the-preview-of-apache-hbase-clusters-inside-microsoft-azure-hdinsight.aspx Enis On Fri, Jul 11, 2014 at 10:01 AM, Nick Dimiduk <[email protected]> wrote: > I believe "transaction log" == WAL. > > > On Fri, Jul 11, 2014 at 9:57 AM, Jean-Marc Spaggiari < > [email protected]> wrote: > > > Interesting. What do they mean by "HBase transaction log files"? Do they > > talk about a transactional framework? Or about the WALs/HFiles??? > > > > > > 2014-07-11 12:22 GMT-04:00 Nick Dimiduk <[email protected]>: > > > > > FYI. It looks like the Microsoft folks want to make Azue the definitive > > > cloud on which to run HBase. > > > > > > -n > > > > > > https://issues.apache.org/jira/browse/HADOOP-10809 > > > > > >
