[Asterisk-Users] Hypothetical : Working across multiple servers??

[Chris Albertson]

The easiest way to handle redundency and fail over might be to
use a level three switch.  This switch can be programmed to
to detect a failed server and then in effect move the ip address
to a working box.  Skipping lots of details you's set up
at least two asterisk boxes each one with two IP  addresses.
say x.x.x.01 and x.x.x.02 in the normal case the switch sends
all the IP packets with ...01 to server A and those with ...02 to
server B but if (say) server B fails then the switch sends
it all to the good server.  Both servers are configured identically.

You can also cross strap these switches so they will fail over at the
hardware level.

The dot com (that went dot bomb) that I used to work at has a
requirement that I should be able to pull a data cable off of
any runnding system's hard disk drive and the end users should
never notice.  Also I should pull any AC power cord without the
end users noticing.  It can be done.  It was really fun testing
too.  We'd load up a test system get Oracle and a bunch of
other servers bashing away and then I'd open up a case and
remove a disk drive, no thses where NOT hot swap.  Just pain old
IDE drives.  Removing one caused all number of errors to be logged
but the end user would not notice.  I has the UNIX sys admin for
this system and selected and built up the system.  We were going to
go Linux/PC but then Sun came out with the $995.00 "Netra" and
just blew away the PCs. The Netra is 1U tall and 10 boxes fit in
less than half a rack. Ten 64-bit SPARCs for $10K what a deal.



--- Steven Critchfield <critch at basesys.com> wrote:
> On Wed, 2003-10-08 at 10:52, WipeOut wrote:
> > Hypothetical question..
> > 
> > Lets say there is a situation where you are using the highest 
> > compression codecs for all extensions (I guess that would be G.729)
> and 
> > the load on a single server is overpowering the most powerful
> single 
> > processor(lets say SMP is not an option).. So two or more servers
> are 
> > required..
> > 
> > Or
> > 
> > The situation is that you need fault tolerance so want to have two 
> > Asterisk servers serving the same population of users and all users
> are 
> > able to authenticate to both servers (I guess you simply copy the
> same 
> > sip.conf to both servers for this) but you also want all other 
> > extensions to be availible to all users irrespective of which
> server 
> > they have connected to..
> > 
> > I guess in a way it would be like an Asterisk cluster of some
> sort..
> > 
> > What would the solution be? Has anyone got an install similar that
> they 
> > would like to share how it was done?
> 
> Wouldn't this be the perfect case for a switch statement, especially
> if
> re-invites worked reliably for SIP? You could then have one machine
> that
> accepted authentication and directed call management via it's
> extensions.conf file. The other machines would then do the
> negotiation
> with the endpoint and then do the codec work also.  
> -- 
> Steven Critchfield  <critch at basesys.com>
> 
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users


=====
Chris Albertson
  Home:   310-376-1029  chrisalbertson90278 at yahoo.com
  Cell:   310-990-7550
  Office: 310-336-5189  Christopher.J.Albertson at aero.org
  KG6OMK

__________________________________
Do you Yahoo!?
The New Yahoo! Shopping - with improved product search
http://shopping.yahoo.com