Re: [Gluster-devel] RFC - "Connection Groups" concept

[Justin Clift]

On 26/06/2013, at 4:42 PM, Joe Julian wrote:
> There are only two translators that use the network, server and client. I'm 
> unclear how these communication groups would get applied.

Yeah, how this would affect things definitely needs to be worked out. :)


> I lean a little bit toward being against solving network problems with 
> application complexity. Can't these problems be solved with split horizon DNS 
> and/or static routing?

Interesting.  Static routing might be a workable approach in some places. :)

Split horizon DNS would probably be possible in some places too, but
nowhere near as many.  Most of the corporate environments I've worked in
(mostly banks/finance, telco, utilities) would likely regard split horizon
as exotic.

Involving the network team in some places can be very, very, very painful,
even for somewhat simple stuff like add static routes added to things. :(

For places like that, if Gluster can do it fairly easily itself (no special
network config needed) that might "open doors" so to speak.

Btw, there's a thread by Kaushal just before mine "How do we identify
peers?", which has bearing on this too.  (and needs to be solved somehow,
though I think they could be solved together)

I guess I should learn more about how Gluster gets implemented in the real
world using split horizon DNS and static routing.  My dev/test lab isn't
big enough to try it one bare metal, so I'll probably have to improvise a
bit. :)

+ Justin

--
Open Source and Standards @ Red Hat

twitter.com/realjustinclift