Archive 19 April 2006 #51
Samples of Redundancy
Applied to VoIP Servers
By William A. Flanagan, Publisher
As a Linux fan for some years, I was encouraged to learn that the high-availability project for Linux (www.linux-ha.org) is doing well--they report thousands of production sites and dozens of equipment manufacturers using it. Release 2 is due soon.
The stable release (L-HA 1.1) allows multiple servers to monitor each other, via an exchange of heartbeat or keep-alive messages--to that extend, very much like Virtual Redundant Router Protocol. Under Linux-HA, if any server dies, or becomes unreachable, a process called Cluster Resource Manager assigns the lost processes on the missing server to another server in the cluster. There are resources in CRM to start the necessary processes on any server. To avoid the problem of porting threads, the project builds instead on The Main Event Loop, a Gnome process that manages all available sources of events (www.gnome.org).
An example of Linux-HA popped up at a recent analyst conference hosted by by Inter-Tel, Inc., a maker of PBXs and voice over IP (VoIP) systems (www.inter-tel.com). Their new generation of call processing for telephony is based on a soft switch--signaling software running on an Intel server. For all larger installations they recommend redundancy, which is based on Linux-HA.
The Inter-Tel people have confidence in the availability and reliability of the platform. They include it in the managed telephony/voice service they offer on a turn-key basis to businesses. They sometimes joke about eating their own cooking--using the equipment they manufacture to deliver a service for which they take complete responsibility. (Inter-Tel has a subsidiary that is a registered CLEC, so it can provide local access as well as the PBX and phones, and they don't sell their lease financing to a third party--they offer "one neck to choke.") They rely on Linux-HA to keep the calls moving.
L-HA appears in other communications equipment, to maintain the control functions for hardware based switches, such as ADC. Check the Linux-HA web site for details and more user stories.
Want to get into high availability without a lot of risk? Try setting up a pair of servers with a less-than-critical function such as secondary or backup DNS server. Tie two Linux platforms together with Linux-HA; most Linux distributions come with the BIND DNS server (the latest versions fix an earlier vulnerability). They will work for you, but won't kill your network if they go off-line.
For just messing around, set up a web site with Apache on Linux-HA. Put a low-priority project on it, to start, and monitor the availability. When your confidence rises, put on more pages.
Let me know what you find.
[In case you haven't noticed, this email is designed for
fixed pitch fonts like Courier.]
"Flanagan Consulting" and "ViewsLetter" are Service Marks of W. A. Flanagan, Inc.
W. A. Flanagan, Inc.
45472 Holiday Drive, #3
Sterling, VA 20166
|In Converged Networking,
We have the Experience