Let me clarify that our analysis involves looking at the changes over a period of time and HBase provides helps us with time based lookups and state of the things at particular point of time.
On Fri, Aug 14, 2015 at 5:16 PM, Jean-Marc Spaggiari < jean-m...@spaggiari.org> wrote: > But why do you want that in HBase? Do you need random read and random write > milliseconds access to those rows? Or you are going to run some analytics > jobs which are going to do full scans anyway? If the later, why not just > dumping that as files and doing the queries with a SQL engine? > > If you need HBase for user, then you might have to build a tool to migrate > the data from MySQL to HBase because the MySQL schema will most probably > not fit the HBase tables... > > 2015-08-14 20:12 GMT-04:00 Buntu Dev <buntu...@gmail.com>: > > > We got a bunch of datasets in our CDH cluster and want to do some data > > analysis on those which requires some meta data only available in MySQL. > We > > could do a one time Sqoop and then want to setup a job to capture the > > changes and write to HBase. I'm looking for options to handle the MySQL > > changes, thanks! > > > > On Fri, Aug 14, 2015 at 5:00 PM, Jean-Marc Spaggiari < > > jean-m...@spaggiari.org> wrote: > > > > > Hi, > > > > > > Before even going into that direction, why do you want to do that? It's > > > most probably not a good idea. > > > > > > Is is for backup? For replication? etc. > > > > > > JM > > > > > > 2015-08-14 19:56 GMT-04:00 Buntu Dev <buntu...@gmail.com>: > > > > > > > I'm looking for ways to setup an incremental update task to replicate > > the > > > > MySQL table to HBase. Sqoop seems like an option to import but > doesn't > > > seem > > > > to handle row updates and deletes. > > > > > > > > Hoping there might be other better ways to handle the > > > > inserts/updates/deletes. > > > > > > > > Thanks! > > > > > > > > > >
