March 1, 2005
Step 5
Service Provider Relationships
Service providers will sometimes provide Service Level Agreements (SLA).
These can be helpful; however, it is essential to measure service level using
metrics that relate to Voice over IP performance. It is important to understand
where the SLA is measured and how it is measured. Problems such as jitter
and packet loss can often occur in access links, e.g., your T1. If the SLA
is measured at the service provider end of the access link, then SLA metrics
do not guarantee good performance.
Since packet loss and jitter typically occur in short periods of time, i.e.,
1-2 seconds, (due to congestion caused by data traffic), then measuring long
term packet loss can be less useful than first thought.
A SLA that commits to less than 0.1% packet loss over a month may sound great;
however, this does not take the effects of jitter into account. Jitter can
lead to packet discards (which are equivalent to packet loss) and can be a
bigger problem on IP networks than packet loss.
This also does not represent the effects of packet loss burstiness. For example,
if there are two bursts, 2 seconds long of 20% packet loss during a 3 minute
call, then the average packet loss rate for the call is 0.44%. This would
lead to the user hearing two periods of severe degradation during the call.
If one call in five experienced problems of this type, then the average packet
loss rate would be less than 0.1%.
New SLA agreements are being developed specifically for Voice over IP. They
address the issues of packet loss distribution and the effects of jitter,
and often represent the SLA in terms of call quality metrics.
Step 6
Define Performance Management Architecture and Tools
A well-defined fault and performance management architecture is essential
for successful network operation, and it should be defined prior to the procurement
of any VoIP equipment. Figure 2 below shows the industry preferred
VoIP performance management architecture which is based on RTCP XR (RFC3611)
and related protocols.
Figure 2: The New VoIP Performance
Management Architecture For Enterprise Networks
It is also essential to consider the issue of manageability which encompasses
both the functionality required in IP endpoints and the protocols they use;
plus any potential conflicts between secure protocols, i.e., Secure RTP, and
the access required by management tools.
The sample RFP Requirements Document shown below (Figure 3) provides
guidance on what IP phone and gateway vendors should provide. The document
introduces the concept of a monitored IP endpoint, which supports RTCP XR.
Monitored IP endpoints are essential if you want to be able to monitor from
the network to the user desktop and detect/resolve transient problems.
| Example RFP Requirements
for VoIP Manageability
1 IP Phones
1.1 All IP phones shall support RTP with RTCP SR/RR (IETF RFC3550).
If IP phones implement Secure RTP (IETF RFC3711) then RTCP SR/RR and
XR reports must be transmitted unencrypted.
1.2 Monitored IP
Phones shall support RTCP XR VoIP Metrics (IETF RFC3611). ITU G.107
and ETSI TS 101 329-5 Annex E (VQmon) shall be used to generate call
quality metrics. All parameters of RFC3611 Section 4.7 must be supported.
1.3 Monitored IP
Phones shall support the appropriate signaling based QoS reporting protocol
- H.460.9 Annex B for H.323, draft-johnston-sipping-rtcp-summary-05.txt
(or later draft) for SIP.
2 IP Gateways
2.1 All IP gateways shall support RTP with RTCP SR/RR (IETF RFC3550).
If IP gateways implement Secure RTP (IETF RFC3711) then RTCP SR/RR and
XR reports must be transmitted unencrypted.
.2 Monitored IP
Gateways shall support RTCP XR VoIP Metrics (IETF RFC3611). ITU G.107
and ETSI TS 101 329-5 Annex E (VQmon) shall be used to generate call
quality metrics. All parameters of RFC3611 Section 4.7 must be supported,
in accordance with ITU G.799.1.
2.3 Monitored IP
Gateways shall support the appropriate signaling based QoS reporting
protocol - H.460.9 Annex B for H.323, draft-johnston-sipping-rtcp-summary-05.txt
(or later draft) for SIP or H.248.30 for Megaco.
3 Probes/ Analyzers
3.1 Probes and analyzers shall use ITU G.107 and ETSI TS 101 329-5 Annex
E (VQmon) to generate call quality metrics for packet based non-intrusive
or active monitoring. 3.2 Probes and analyzers shall support the detection
and analysis of RTCP SR/XR (IETF RFC3550) and RTCP XR (IETF RFC3611)
VoIP Metrics payloads, including the extraction of parameters related
to delay, signal/noise level and endpoint configuration.
4 Embedded
SLA Monitoring Function in Routers
An Embedded SLA Monitoring Function is a software agent installed in
a service provider managed edge router, multi-service gateway or integrated
access device located on the customer premise. The purpose of the SLA
Monitoring Function is support the measurement of service level received
by the customer, and to permit the service provider to implement some
level of remote diagnostics. This is achieved by gathering statistics
on live customer traffic and by implementing active testing for both
inter-site monitoring and on-demand troubleshooting. There are two service
models that may be supported:
- Managed VPN
Service, in which all traffic is carried on encrypted tunnels between
customer location 7
- Managed VoIP
Service, in which the service provider participates in call management.
Examples include IP Centrex/ Hosted PBX service.
|
Figure 3: Sample RFP Requirements Document
Deploy Network In Stages
VoIP network deployment should be planned and executed in stages—with
an initial pilot trial preceding large scale deployment. And make sure that
the pilot trial includes some typical satellite locations, branch offices
and teleworkers—since these are typical problem area hotspots.
Early during your network deployment, check for (and resolve) the following
problem areas:
- firewalls blocking access to voice traffic
- congested access links leading to high levels of jitter
- echo on incoming telephone network connections, i.e. non-VoIP connections
to the phone company.
Summary
This Tech Note provided a six step methodology for enterprise network managers
to follow when preparing their network for Voice over IP service, including
pre-deployment testing and network readiness assessment. The document also
described many of the key factors essential for successful deployment of VoIP.
Other useful resources include: the VoIP
Troubleshooter Web Site, and application notes on VoIP performance management
available from specialized VoIP performance experts such as Telchemy.
Acronyms
|
|
| CDR — Call
Detail Record |
RED —
Random Early Detection |
| IETF — Internet
Engineering Task Force |
RTP —
Real Time Protocol |
| IP — Internet
Protocol |
SLA —
Service Level Agreement |
| ITU — International
Telecommunications Union |
SNMP —
Simple Network Management Protocol |
| MIB — Management
Information Base |
TDM —
Time Division Multiplexering |
| PBX — Private
Branch Exchange |
VoIP
— Voice Over Internet Protocol |
| POTS — Plain
Old Telephone Service |
VPN —
Virtual Private Network |
| PSTN — Public
Switched Telephone Network |
VQmon/EP
— VQmon End Point |
| QoS — Quality
of Service |
VQmon/SA
— VQmon Stream Analysis |
IE 7
Firefox 2.0