short term; simulated (synthetic) VoIP calls and other traffic are generated to assess suitability for VoIP, VoIP load capacity, testing of call controllers or other VoIP infrastructure; quality of service performance and/or SLA compliance.
A year ago we still cautioned users to make sure any VoIP analysis tool they bought supported SIP,
ideally as a standard component. That is no longer necessary. All of the leading packages now support SIP integrally.
It’s fast becoming a SIP world, and users from now on will have an increasingly hard time finding analyzers that understand their particular vendor’s proprietary protocols although Cisco SCCP (Skinny), Nortel (Unistim), and Avaya
(H.323 variant) are still fairly common.
Active or passive VoIP monitoring remains a bone of contention although most of those on the table now support both capabilities.
Active VoIP call-quality measurement is where simulated (or synthetic) VoIP calls are placed by, and measured between, strategically located test devices. This naturally consumes bandwidth on IP routes between these points; therefore, it is called intrusive monitoring. Ratings usually, but not always, employ one of the perceptual evaluation of speech quality (PESQ) measurements, which compare a received audio file to a reference original. The results generally are in the form
of mean opinion score (MOS) 1-to-5-scale ratings.
Passive monitoring means watching and rating real-user VoIP traffic. Representative of passive call quality assessment is the R-factor metric, known by other names and references including
R value, E Model, and ITU G.107. Component criteria include loudness, noise, vocoder type, and delay. The result is a value on a 0-to-100 scale.
A good-as-it-gets R-factor rating is 93, which can only be achieved with G.711 vocoding and everything else being nearly ideal. The MOS equivalent, by the way, is 4. 4. Similarly, R-factor scores of 80 to 100 map to MOS scores of 4.0 to 5.0; R factors of 60 to 80 equate to MOSs of 3.0 to 4.0. And calls below an R factor of 60, or a MOS of 3.0, are generally considered unacceptable.
Both techniques have their place; however, from the many network analyses it has performed of both enterprise networks and carrier services, MierConsulting has
found that active VoIP monitoring provides a generally better and more controllable, consistent, and accurate assessment of VoIP call quality than passive monitoring of real-user VoIP traffic.
The “point-of-view” issue also impacts passive monitoring. A passive VoIP call-quality rating can depend on where the passive analyzer is situated, usually on a backbone link somewhere near the core of the network. This may not necessarily represent
References:
Archives