[Gluster-devel] Glusterd: A New Hope
anand.avati at gmail.com
Thu Apr 4 20:37:34 UTC 2013
On Sat, Mar 30, 2013 at 1:36 PM, Justin Clift <jclift at redhat.com> wrote:
> On 22/03/2013, at 2:09 PM, Jeff Darcy wrote:
> > The best known example of such a coordination service
> > is Apache's ZooKeeper, but there are others that don't have the
> > noxious Java dependency - e.g. doozer written in Go, Arakoon
> > written in OCaml, ConCoord written in Python.
> Have been idly thinking about this for a few days.
> Gluster already uses both C and Python, so ConCoord sounds like the
> best option to investigate from those.
> Saying that because the possible addition of another language into
> Gluster - and not for trivial bits - really, really, worries me. :(
> We did that with Aeolus (both Python and Ruby), and it was an
> _extremely_ bad idea in hindsight. Every time non-trivial problems
> occurred between a Ruby part and a Python part, we couldn't get
> "just anyone" in the project to fix it.
> We had to get people involved (simultaneously) who know both
> languages well, their debugging tools, etc.
> i.e. Bottleneck / choke point. :(
> It can also be a strong deterrent to getting new features in, if
> they touch "the other language" part in non-trivial ways... again,
> not just anyone can do it.
> So, even if ZooKeeper/doozer/Arakoon are wonderful... please not
> them. :)
> Regards and best wishes,
> Justin Clift
Justin, thanks for the comments. I agree with this point, that we should
pick a solution which has widest coverage of programming language
familiarity within the team. I think Python and C are the leading languages
in the current team. I am yet to look into the specific project suggestions
made across the thread, hopefully there are some good Python based ones.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Gluster-devel