Home arrow Articles arrow System Reliability
System Reliability PDF Print E-mail
Written by Michael Keilhofer   

Reliability is a measure of how confident we are that an attempted call will be successful. Reliability can be addressed by reviewing three major areas:

  1. The hardware.
  2. The software.
  3. The network.

Redundancy and high-availability are the two terms most often used to describe how reliability is improved. Use of either of these can result in increased reliability.


Most customers, or end users:

  • have some control over network reliability.
  • may have complete, limited, or no control over hardware reliability.
  • have little or no control over software reliability.


To improve reliability we typically focus our attention on critical (most prone to failure) compenents.

Redundancy typically refres to duplication of critical components such that if one fails the other will continue or take over and result in no or minimal impact of call processing.

High availability typically means that critical components are either made better or redundant.


To address high availability systems one can review the following areas:

  • Install a much larger power supply than needed. This will result in a much lower chance of the power supply failing because it is working much less to supply the power needs of the system.
  • Use a Server motherboard rather than a desktop motherboard. Server boards are not only made better but the air flow is correct for the front to back normally applied to rack mount erquipment whereas a desktop motherboar's components are mounted in a way to constrict front to back air flow.
  • Use a better disk drive. Every disk drive has n MTBF (Mean Time Between Failure) rating. Many drives also include a waurante that is longer than other. SCSI drives are, in general, always built for a heavier load and will last longer. The exception to this is the Western Digital Ratpor SATA drives which are considered tough enough for IT equipment.


To address redundancy system one can the following things:

  • Install your disks in a RAID 1 or RAID 5 configuration. This means if one of the disks fail the system will not stop. Using hot swapable drive carriers also minimizes down time by allowing the replacement of a failed drive while the system is running.
  • Slect a chassis that has dual or even tripple power supplies. Then if on eof the power supplies fails, it too can be replced without the system being stopped. Consider going a step farther and even having the AC (or DC) power feed to each power supply come from a different source.
  • Make sure you have dualEthernet connections on your mother board and use them on a redundant LAN.

To discuss reliability with regards to a network we should understand the components that make up a SIP network.

A SIP-based network is comprised of several basic components:

  • SIP User AgentA SIP enabled end-point that can originate or terminate a SIP session. This includes a SIP enabled telephone, a PC based softphone, or a SIP-enabled Gateway.
  • SIP Proxy ServerA SIP proxy server receives SIP messages from user agents and acts on their behalf forwarding or responding to those messages. It is used to authenticate and authorize users for services, implement provider call-routing policies, and provide features to users
  • SIP Redirect ServerA User Agent Server that provides routing information to user agents, giving the user agent an alternate uniform resource identifier (URI) or other user agent server (UAS).
  • SIP Registrar ServerA user agent server that receives REGISTER messages and maintains a directory of logical location of each registered user agents within that domain or sub-domain.

Each of these can be installed on a different server and accessed via a redundant LAN. In this way, true redundancy can be achieved to allow maximum reliability.


Last Updated ( Saturday, 02 September 2006 )