By reducing the complexity of distributed systems through separation, and defining formal operational
work-flows, the reliability of VoIP services can be improved heavily.
The existence of popular Open Source Voice-over-IP (VoIP) frameworks based on the Session
Initiation Protocol (SIP) allows service providers to easily deploy feature-rich telecommunication
systems. However, despite the rapidly growing spread of VoIP services, a lot of providers still face
problems when trying to establish a high level of reliability of these services.
In order to determine potential causes for availability degradation, traditional telephony and Internet
services as reference models for distributed communication systems are analyzed according to
their reliability and scalability. Since human errors are found to account for the majority of system
outages, guidelines for designing and operating a complex distributed environment are defined.
Also the usage of SIP in typical network topologies raises problems when it comes to scaling a
VoIP service. Therefore a practical approach for providing a highly modular and thus scalable
system is presented.
Copyright (c) 2005-2008 by Andreas Granig