Redundancy

From Pbxnsip Wiki

Purpose

The PBX is a critical component in the office infrastructure. For many offices, having no PBX service running in the network means that business is affected. Because of this, the there is a need to prepare for the situation when the service fails.

In the case when the software fails, the operating system will take care about a restart of the service. In Windows, this is done by the service manager. We recommend automatically restarting the service (see Installing in Windows).

Manual Failover

Hardware failures are much more difficult to recover. There are a couple of things that you should do in order to prepare for a hardware failure:

• After having the installation done, make a backup of the working directory in a place that you can easily access (e.g. a USB memory stick). This backup will make your life easier in any case. Also consider cases like fire or theft of server hardware.

• It does not hurt to make a backup from time to time. Users change settings, and you also might create new accounts and record new announcements. If you have a backup, you can easily move back to this point later.

• Also consider other central network components to fail. It might happen that your server works fine, but your LAN switch breaks. Every hardware component can break.

• Think about what you will do when the server breaks. Maybe you have another server that you can use in this case, even if the service will not run as good as on the primary server. Check the license type. If you have a USB license key, you can easily move that key from one server to another. If you have a MAC-based license key consider getting another key for the secondary server.

When you bring up another server, it is good if you can use the IP addresses of the primary server. Unless you use DNS on the registered devices to locate the SIP server, it makes your life much easier if you just re-use the same IP address that you were using on the server that failed.

Automatic Failover

If the manual failover is not an option for you, consider an automatic failover. The automatic failover does nothing else than what you would do manually. The only difference is that you let a program do the necessary steps.

If you run an incremental file system backup from the primary server to the secondary server every few minutes your database should be reasonably in sync with the secondary server.

In order to be able to change the IP address automatically, you need some kind of heartbeat between the primary and the secondary server. There are products on the market that do this automatically for you, for example the heartbeat project. Then in the case of a failure, that program needs to start the PBX service on the secondary server and keep the primary server down when it comes up again.

If you run a server farm with more than one primary active server, you may share one secondary server with several primary servers. This server will cover your server farm unless more than one server fails before you can bring another secondary server online.