Amit, 

That sounds like a reasonable idea. You should be careful about handling SQL 
related issues (like duplicate primary keys etc). You'd probably need to add a 
lot of the SQL related configuration and failure handling 


Thanks,
Hari


On Saturday, October 12, 2013 at 8:48 AM, Amit Sela wrote:

> Well, like I mentioned, I'm still new to Flume so maybe I'm totally of course 
> here, but how about an implementation similar to HBaseSink ? replace the list 
> of Puts with a list of Strings of insert statements ? execute them in batches 
> ? 
> 
> 
> On Thu, Oct 10, 2013 at 7:32 PM, Steve Morin <[email protected] 
> (mailto:[email protected])> wrote:
> > I would be interested in knowing your use-case too
> > 
> > 
> > On Thu, Oct 10, 2013 at 8:47 AM, Jarek Jarcec Cecho <[email protected] 
> > (mailto:[email protected])> wrote:
> > > Hi Amit,
> > > to my best knowledge, there is no MySQL sink that would store data from 
> > > Flume in MySQL. Would you mind describing your use case? I would be 
> > > interesting to hear how the MySQL fits into the use case. Flume is heavy 
> > > distributed system for collecting and injesting data, that can very 
> > > easily overload single MySQL server, so I'm wondering if you are planning 
> > > to have some sort of sharding or perhaps a mysql cluster to sustain the 
> > > load?
> > > 
> > > Jarcec
> > > 
> > > On Thu, Oct 10, 2013 at 09:10:34AM +0200, Amit Sela wrote:
> > > > Hi all,
> > > >
> > > > I'm new to Flume and I would like to know if anyone used it to aggregate
> > > > logs and process them into MySQL (JDBC) inserts ?
> > > >
> > > > Thanks.
> > 
> 

Reply via email to