Switching to VoIP

This is the Title of the Book, eMatter Edition

Copyright © 2007 O’Reilly & Associates, Inc. All rights reserved.

Interoperability

|

261

To resolve the simulated incompatibility, remove the md5secret token from sip.conf

and attempt to place a call with the X-Lite softphone again.

Troubleshooting Quality-of-Service Issues

Not all troubleshooting situations involve signaling. As you may have read in prior

chapters, noise problems and dropouts stemming from jitter and packet loss are

cause for troubleshooting. Fortunately, the standards for sound transmission—RTP

and codecs—are simple compared to call signaling, so troubleshooting sound issues

is generally easier.

That’s not to say all sound-related issues are a snap to dismiss. Sound issues are

affected by many disciplines of networking: switching, routing, QoS, and packetiza-

tion. The way congestion on a routed link shows up is in dropouts, or in busy sig-

nals if you’re using a QoS solution. On a switched link, like Ethernet, busy signals

may occur long before dropouts are perceptible. So symptoms are different on differ-

ent links.

As with signaling, good diagnostic logging is valuable for troubleshooting. Packet

analysis is less useful when examining sound quality problems. One can certainly

identify jitter using packet capture, but it’s just easier to identify jitter across IP links

using tools like traceroute and pathping, which succinctly show the variances in

round-trip latency on every hop of a route. If you need to dig deeper, capture tools

like Ethereal and its commercial equivalents can assist you in isolating jitter at the

RTP level.

Commercial packet analysis tools

If firewalls or other filters block ICMP or TCP traffic along the call path you’re trou-

bleshooting, then you may use a packet capture tool with advanced VoIP capabili-

ties, like Network General’s ubiquitous Sniffer or WildPacket’s EtherPeek VX. Both

permit analysis of signaling and media packets, readily understand the structure of

RTP, can identify jitter, and can reveal sources of congestion on your network.

When the going gets tough, call in the end users

The human ear is a great instrument for troubleshooting. You can carry an IP phone

from one server closet to the next, listening to calls to the same destination from dif-

ferent points along the call path. This process of elimination will help you narrow

down a group of potential congestion points until you’ve identified the one that’s to

blame. On large, expansive networks, where this can’t be done by a single person,

you could enlist help from other people within the organization. Even end users are

pretty reliable when it comes to assisting with this type of troubleshooting. Ask them

to make a call that you expect to traverse the call path you’re looking at, and then

ask them how it sounds or if they get a busy signal. Both can be indicators of conges-

tion, though a busy signal will happen only in networks with QoS.