Failover between two SIP External Gateways

[Kasper-1973]

Hi all,

In the Cisco world, I could configure two dial-peers with preference 1 and 2. They could both be configured with SIP keep-alive, so that if the first one is unreachable, the second one is used instead.


How is this done in the Alcatel world?

Does it have to be via Professional Trunk Group Seizure and ARS or is there a simpler way to do it?

Kind regards
Kasper

[tot3nkopf]

Depends...on what you need and on the type of external gw.
=>You need also round robin distribution of calls or only overflow? You can use the sip ext gateways pool mechanism for the first one.
=>You can use ARS as you mentioned for overflow. The behaviour depends on software release and implemented configuration.
=>You can use DNS SRV or DNS A with two IP addresses.

Check the alcatel system doc for details.

[Kasper-1973]

Hi tot3nkopf
Thanks a lot for the swift reply.
Software version is 11.1

You need also round robin distribution of calls or only overflow? You can use the sip ext gateways pool mechanism for the first one.

No round robin is needed - It's simply for failover.

You can use ARS as you mentioned for overflow. The behaviour depends on software release and implemented configuration.

I believe that this is a good solution - It's not easy to implement because the number plan is messed up, but it's possible.

You can use DNS SRV or DNS A with two IP addresses.

I will take a look at this, 'cause it makes sense to use DNS, I just don't know enough about it yet. I'm sure that the system docs will help me.

I will be back in a month or two with the solution, when I have access to the equipment again.
again - Thanks a lot so far.

/Kasper

[Kasper-1973]

Hi tot3nkopf,

I did some testing on the systems and tried to get the DNS SRV/DNS A up and running, but without luck. I could see that the DNS reply included two IP addresses, but for some reason the OmniPCX chose to use the first IP only. So that didn't work out.

Then I turned to the SIP Ext. GW pool mechanism - That worked out quiet good and solved the problem. I'm just no sure how this will solve the issues in the future, since I only have pools 1 to 4 to use from. But I will cross that bridge when I get to it.

I will try to mess around with the ARS when I have more time.

Thanks a lot for your time.

Kind regards
Kasper

[tot3nkopf]

Hi,

For the DNS stuff how does the system behave when the first IP address returned by the DNS server is not reachable? I have no experience with this however my assumption is that only when the first IP is not available it should use the second...

Rgds

[Kasper-1973]

Hi,
Sorry for the late answer.
I have done some research and testing during x-mas and here are my results.

Tested scenario:
In the first scenario the SIP Ext. GW. is configure with a name and one DNS Server. The DNS Server is a standard windows server configured to respond with two IP addresses.
1. The Call Server sends a request to the DNS Server.
2. The DNS Server responds with two IP addresses.
3. The Call Server tries to establish a call using the first IP address, which is the IP address of one of the SIP Servers.
4. The SIP Server does not answer.
5. The call fails.

NOT tested scenario - The reason for not testing this is that I can't test it when the design includes a Passive Call Server. In order to test this scenario, I would need a redundant setup e.g. Spartial Redundancy.
The idea is to use the Main Call Server as primary DNS Server and the Redundant CPU as the secondary DNS Server.
The Call Server has been configured with a SIP Ext. GW. with a name and two DNS Servers.
1a. The Call Server sends a request to the first DNS Server.
2a. The DNS Server responds with an IP address.
3a. The Call Server tries to establish a call using the IP address, which is the IP address of one of the SIP Servers.
4a. The SIP Server answers.
5a. The call is established.

1b. The Call Server sends a request to the first DNS Server.
2b. The DNS Server does not respond with an IP address.
3b. The Call Server sends a request to the secondary DNS Server.
4b. The DNS Server responds with an IP address.
5b. The Call Server tries to establish a call using the IP address, which is the IP address of the redundant CPU.
6b. The SIP Server answers.
7b. The call is established.

I have attached the pdf which explains the second NOT-TESTED setup.

Spartial Redundancy.pdf

Kind regards
Kasper

[tot3nkopf]

Which one is your case (architecture) and what are you still not able to achieve?

[Kasper-1973]

Hi,

I have attached my setup.

Architecture.pdf

I have achieved what I wanted to in the first place. (Monitoring of a SIP Ext. GW)

I will receive a new set of hardware with-in the next month or so and then I will try to do the DNS setup again. I just don't understand why the documentation says that the NODENAME must be unique when it want's me to configure "brest" as nodename on both shelfs.

I will let you know the results.

/K

[tot3nkopf]

I think this was discussed in the forum before.

Check these:
viewtopic.php?f=264&t=24334&hilit=sip+dns
viewtopic.php?f=264&t=23988&p=86490&hil ... dns#p86490

Node name needs to be unique: it is the same node. We are not talking about shelves but about CPUs.

[tot3nkopf]

By the way are the two nodes networked using ABCF?