If you are a network administrator, you've heard the mantra "redundancy." Greatly simplified, that means complete-as-possible, real-time replication of data across different paths. If one path goes out, well, you have Plan B. The goal, especially in data networks, would be to have a "five 9's reliability" that would cause you network to be down only 10 minutes a year.
A worthy goal, yes - but research and consultancy firm Gartner is out with brief today that may cause sysadmins that run VoIP to reassess exactly what "redundancy" means. The brief is entitled "How IT Managers Can Make VoIP Networks More Reliable."
"Unfortunately, 'five 9s' reliability' is difficult to achieve in data networks," writes report author and Gartner consultant Jeff Snyder. "Complete redundancy would certainly increase the reliability of a network, but is usually cost-prohibitive; (thus) the degree of reliability 'appropriate' to a company, therefore, usually becomes a financial decision.
Snyder recommends some redundancy policies, including:
- Redundant servers,co-resident or hosted in remote locations;
- Backup servers located in regional offices;
- Redundant network equipment with multiple access paths.
Beyond redundancy, Snyder endorses the following IT policies for VOIP network management:
- Keeping "a limited number" of phones around, connected to the Public Switched Telephone Network;
- Because lines can be cut, keep a physically independent alternate route open to your Network Service Provider;
- Use your PSTN as an emergency backup to your Wide Area Network;
- Identify specific phones and locations in you network with the highest need for emergency power backup, and then implement that solution.