atlas-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Seetharam Venkatesh <>
Subject Re: Atlas & Storage back-ends
Date Tue, 19 Jan 2016 20:30:36 GMT

Since most of the active contributors are very familiar with HBase, we have
tested it with HBase and documented the steps in the wiki. We have run any
formal performance but we know HBase scales.

OTOH, if anyone from the community is willing to test Cassandra, we
definitely would love to help them.

Also, Titan folks have done extensive testing and published numbers for
both Hbase and Cassandra. The workloads might be different but we are using
Titan as the graph DB.

Makes sense?



On Fri, Jan 15, 2016 at 4:03 PM Hemanth Yamijala <> wrote:

> Nigel,
> There has been no comparison done so far between HBase and Cassandra. That
> said, we are very much focussed on getting some work done for evaluating
> the performance and scalability of Atlas overall - hopefully very soon. As
> part of that we would certainly evaluate how best to tune HBase for optimal
> performance with Atlas and publish those findings.
> Hope that helps a bit.
> Thanks
> hemanth
> On Wed, Jan 13, 2016 at 7:41 PM, Nigel Jones <> wrote:
> > Has anyone deployed atlas "in anger" ie to support any kind of production
> > usage?
> >
> > If so have you done any comparison on Hbase vs Cassandra as the storage
> > backend? I know there was a jira relating to cassandra configuration but
> > that was resolved.
> >
> > My gut feel is Cassandra could be a more scalable option, but having not
> > done any testing myself (yet?) I'm wondering how scalable the Hbase
> > approach is today, and which might be a safer, more manageable option.
> >
> > Nigel.
> >
> >

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message