[Freeswitch-users] HA questions.
freeswitch-users-list at metik.com
Fri Dec 4 07:24:36 PST 2009
Since you seem to have most of the heavy lifting squared away with FS
(e.g. database replication) and before reinventing the wheel, I would
recommend that you speak to a few VoIP providers and see if they will do
this for you as part of your service. Those that are using carrier
class platforms (so-called active clustering) should be able to do this
without too much effort on their part. If you reach any dead ends,
please feel free to contact me off list.
The one thing that you may want to keep in mind is that unless FS is not
involved in the media flow (or has a chance to redirect the call to
another FS), existing calls will be dropped. FS has no mechanism for
exchanging/mirroring stateful signaling and media information between
other FS nodes to specifically facilitate failover. As I believe the
developers have indicated in this list, to do so would require
significant investments in time and resources to implement it in the
sofia sip stack at the moment.
Tim Uckun wrote:
> I have read some of the archived emails about HA, loadbalancing,
> failover etc and I am still a bit confused about how I could set up
> some sort of resiliency with freeswitch.
> My situation is much less complex than the scenarios people were
> talking about and I hoping the solution is similarly much less
> I have two machines. Both will run freeswitch and also an IVR
> application with local databases. I will take care of the database,
> application and configuration synchronization between the two
> machines. Ideally the calls would be load balanced between the
> machines and if any application falls down then the calls should go to
> the other machine. Same if I take a machine down for whatever reason.
> If a machine goes down I am willing to "lose" those people who were
> making a call at the time. I do have a flag in the application which
> will stop answering the calls while processing the existing calls for
> a graceful shutdown and hopefully the load balancer would shuttle the
> calls to the other machine while this is happening.
> At this stage everything is done via SIP.
> My questions are...
> Do I have to have a sip proxy? If the answer is yes it seems like I
> have to set up two sip proxies so I don't have another single point of
> failure. Can I load the sip proxies on the same machine? Do I need two
> more machines?
> If I take load balancing out of the picture would it be possible to do
> a simple linux HA or a windows built in ip failover solution? Would a
> simple IP failover work over UDP or would I have to use IAX and tcp/ip
> Is it better to go the virtualization route?
> Sorry if these are dumb questions. I am just trying to get my head
> wrapped around this. I don't need five nines (although that would be
> awesome), I just want a reasonable degree of assurance that my app can
> keep taking calls in case something weird happens.
> FreeSWITCH-users mailing list
> FreeSWITCH-users at lists.freeswitch.org
More information about the FreeSWITCH-users