VoIP without monitoring is like cooking without tasting.

Post a comment

by Brian Boyko
Editor, Network Performance Daily

I once made stew. Once.

As it turns out, I'm a horrible cook. Partially because my thought process was something along the line of: "Hmm… we don't seem to have any browning sauce. What do we have that's brown?"

As it turns out, soy sauce? Not so good in a stew.

I made enough of this stuff for three people, and inflicted it on my roommates. Let's just say that despite the heat being over 98 degrees in Texas, the atmosphere in the apartment was much cooler for a little while…

We all cook our own meals separately now.

You know what, I probably could have avoided that if I had actually tasted the food before I served it to my roommates.

Now watch as I take that almost totally unrelated anecdote and tie it into the NetQoS VoIP Monitor launch today. Abracadabra!

(continued…)

Networks send bits. That's the thing to remember about VoIP - there's no actual "sound" coming across those wires, though it's easy to think this way. The voice is digitized. (Of course, this is true for PTSN as well, for a while, conventional phones digitize analog signals as well. What's different is that network with VoIP are converged, voice and data.) But there's more to a phone call's quality than just tracking the bits. Things like delay and jitter become much more important. Packet loss can become annoying or even disruptive. And most network monitoring products really aren't optimized for looking for the things that specifically affect VoIP quality. The data can "look good" but, in reality, it doesn't matter if the bits are still getting to the destination without clogging up the link if the people speaking to each other simply can't be understood!

It's a bit like thinking that soy sauce makes a good browning sauce. Sure, the stew looked brown. It looked like really good stew. It tasted like crap.

We just released the NetQoS VoIP Monitor product at the VoiceCon show in San Francisco. It monitors VoIP call setup and call quality on IP telephony systems. It tells network and VoIP managers how network performance affects the quality of experience for users of VoIP phones. We're proud of it, and we think that it answers the very specific concerns of those rolling out, maintaining, or looking to improve the quality of their IP telephony.

The NetQoS VoIP Monitor augments the other products in the NetQoS Performance Center, which measure VoIP traffic flows and monitor the IP telephony infrastructure, by providing real-time and historical views into call setup (time to dial tone, time to connect after last digit dialed), call quality (MOS score), and the network metrics that can impair VoIP quality of experience. And it allows you to monitor in-flight calls for specific phones that are problematic or critical to your business.

You can take steps to ensure network performance, but without actually measuring call quality experience, it's like cooking a recipe without actually tasting the food. Just because it looks good, doesn't mean it is good.

What NetQoS VoIP Monitor does is monitor the call quality constantly, tracking packet loss, jitter, delay, the sort of things that may be overlooked or downplayed when monitoring data apps, but which are actually crucial to the call quality experience. The NetQoS VoIP Monitor uses a network-centric approach to VoIP monitoring, instead of interrogating the IP PBX database for call metrics, and is packaged as a drop-in appliance (there are single unit and distributed configurations depending on the IP telephony network). It connects to the SPAN port on the switch nearest to the IP PBX (Cisco CallManager) and requires no software agents to be deployed in the IP telephony infrastructure.

We've covered VoIP extensively in the past - and surely will some more in the future. In our Whiteboard Series, Jim McQuaid talked a little bit about the difference between UDP and TCP traffic, and how UDP can dominate TCP traffic if you're not careful. I wrote a bit about what the Mean Opinion Score means and why it, and not latency or data transfer rate, is the most important metric in VoIP monitoring.

That's not just because VoIP is a hot topic, but because VoIP is a difficult technology to get exactly right. Skype, for example, a company whose entire business is based on bringing VoIP to the masses, went down for two days over an unforeseen bug and a Microsoft Update. Even the most focused of us on this topic can still have problems.

It may just be that VoIP is one of those technologies which will undoubtedly mature in the long term - but which companies need now. Those can be nightmares for network engineers and other IT professionals.



TrackBack

TrackBack URL for this entry:
http://www.netqos.com/MT/mt-tb.cgi/285

thepipe-sidebar.png

networkobservations-sidebar.png