Howto:SIP redundancy / failover

FIXME: this article is still in progress and not finished yet

Overview

Often scenarios are needed in which a reliability of the SIP connection is to be realized. There are several approaches and scenarios for SIP redundancy or failover.

How well SIP redundancy and / or failover works depends on the interaction between the PBX and the carrier providing the SIP account.

PBX / Gateway

To support various carriers, we bring natively the following features which can be configured:

• Accepting trunk registrations from multiples Sources (GWs)
• Always place the call on a primary trunk and in case of a fault on a secondary trunk or round robin
• Receive calls from multiple trunk registrations
• Route calls between GWs back and forth
• Rerouting of failed calls

Carrier scenarios

Now it depends on the carrier featureset what you can do

Two permanent active SIP Interfaces

The best way if the Carrier accept two registration from different souces with the same credentials

• SIP Interfaces on two Gateways with equal configuration
  • Both Interfaces register to the Trunk Object in the PBX
• Routing on the Gateway should be equal
• The call distribution is not critical, only needs to be considered in the configuration
  • Some carriers put al incoming calls only on the primary trunk and expect all outgoing calls over the same primary trunk
    • Outgoing Calls must first send to the primary Interface, then to the secondary one
  • Some carriers use both trunks for incoming and outgoing calls

Active SIP Interface, secondary disabled SIP Interface

If the Carrier doesnt accept two registration from different souces with the same credentials cou can prepare the failover, but you have to switch manually.

• SIP Interfaces on two Gateways with equal configuration
  • The primary Interface should be enabled
  • The secondary Interface should be disabled
  • Both Interfaces register to the Trunk Object in the PBX
• Routing on the Gateway should be equal
• The call distribution is straight forward because ther is only a single active Interface at the same time
• If the primary trunk fails, it must be disabled manually and the secondary activated