High Availability (HA) is normally achieved through "clustering" - which means two machines acting as one for a specific purpose. There are many ways to create a cluster, each with its own benefits, risks, costs, and trade-offs. The terms "High Availability" (HA) and "Clustering" can be overused so beware of the hype. Clustering, and HA have specific (and different!) meanings. If you are responsible for creating a high availability cluster for Asterisk, below are the issues and concepts you should be aware of. This page is intended to be a starting point in the design, creation or selection of a High Availability or Clustering solution for Asterisk.
Please do not add specific product names/links to this page, it is intended to be product neutral. Fault Tolerance vs Clustering vs High Availability
Before you get started, you should know the difference between these three things. They all mean different things, and you need to be careful what you refer to.
Fault Tolerance
When a fault is detected, it repairs itself quickly. This must NOT result in an outage (for example, a failed CPU that requires a reboot), but the machine must continue working without an interruption. This is normally achieved with Redundant power supplies, network cards, and RAID.
Clustering
When something that is NOT fault tolerant breaks, you must be able to continue on, on different hardware. This may be a CPU failing that requires a reboot., or a kernel panic. At this point, another node should immediately start the services that were on the failed node.
High Availability
HA is a concept, rather than a thing. HA refers to the combination of both Fault Tolerance
and clustering, as well as network and power design that removes as many single points of failure as possible. You get to imagine running around with an axe, randomly breaking things, and figuring out what to do when that happens.
Co-Dependence and Autonomy
In order to be a true cluster, the machines (or "peers") must share as little as possible. Some HA solutions involve sharing hardware, software, a logical device, etc .The problem with this approach is that you create a single point of failure. For example, if a cluster shares a hardware channel bank (eg: connected to 2 machines via 2 USB cables), then if the channel bank fails the entire cluster fails. ...