SAN DIEGO -- If designed properly, IP telephony technology can provide users with higher availability than traditional, circuit-switched telephone networks, according to one industry analyst.
Speaking at the Burton Group Catalyst Conference Thursday, Eric Siegel, a senior analyst with the Midvale, Utah-based firm, said that an IP network architecture is appropriate for high availability voice applications. But often times, he added, poor implementation of such applications leads to lower than desired levels of availability.
"You really could, surprisingly, get really high availability out of your IP telephony if you set it up properly," Siegel said.
From physical failures in LANs and WANs, to bandwidth congestion and environmental failures, any number of problems can negatively impact IP telephony systems. But Siegel said that with proper redundancy and other best practices, these problems could be avoided.
When it comes to the health of the overall architecture, Siegel said users should look into distributed call control services offered by networking vendors.
"Cisco and others have a kind of rudimentary call control that you can install on a regular router, so that if you lose connectivity with your central office where the main call controller is, that little router in the branch office can take over," he said.
In the area of switching and routing, Siegel said it's very important to protect important data paths by employing redundant common logic and load sharing.
To improve switch re-convergence times, Siegel said enterprises should make use of the IEEE's 802.1w Rapid Spanning Tree Protocol, which restores data paths about 30 seconds faster than its predecessor, Spanning Tree Protocol, 802.1d.
Siegel recommended using standby or load-balancing features to provide backup for the gateway. Also, to avoid the problems that tend to arise with LANs and associated equipment, he endorsed having redundant common logic in important switches.
"Please avoid shared Ethernet," Siegel said. "It's just all too sensitive to crises and messes."
Because faulty cables cause many VoIP problems, Siegel said it's important to avoid using too many small switches.
"Eighty percent of the time it's the cables," he said. "So, let's use fewer cables."
He also suggested to make sure that incoming voice traffic travels on a separate virtual LAN.
"Separate yourself from all the hacking chaos that is going on in the data network and the incoming flood of DDoS attacks," he cautioned.
For extra protection in the area of WANs and associated equipment, Siegel said companies should look into SDH multiplex section protection and automatic protection switching using SONET., the standard for synchronous data transmission on optical media.
To protect physical assets, Siegel said to make sure that access to the wiring closet is highly controlled.
"Lock the thing," he said. "You should be locking it anyway for various reasons."
Conference attendee Kris Kline, an enterprise network architect with Basking Ridge, N.J.-based Avaya Inc., agreed that redundancy is the key to high availability of voice over IP (VoIP).
"Make sure you have redundant capability not only at the edge, but at the core as well," Kline suggested. "(Siegel) was right when he said that availability is really about making sure you have multiple paths to take."
Daniel Fairless, an IT administrator with Oklahoma University's Health Sciences Center in Oklahoma City, said his organization is currently examining a possible switch to VoIP. He had one piece of advice for anyone considering such a move.
"Make sure that the voice traffic is isolated from the data," Fairless said. "This provides more reliable transport and ensures proper security."