Example: A Geographically Distributed Solution - Avaya Application Solutions Deployment Manual

Local Area Network (LAN)
The LAN segments of an enterprise network are typically owned and controlled by the
enterprise. As such, the responsibility for LAN availability rests with the enterprise. With proper
engineering and redundancy planning efforts, Avaya believes that a enterprise's LAN has the
potential of meeting 99.99% to 99.999% availability. Such high availability depends on proper
network design following industry best practices, change control procedures to minimize
operator error, and proper network management techniques to quickly identify and repair
outages.
Wide Area Network (WAN)
A portion of the traffic in a geographically distributed IP Telephony system will traverse the Wide
Area Network (WAN). WAN link failures are one of the leading causes of network outages.
These failures can be due to cable failures, equipment failures, service provider errors,
administrator errors, or malicious activity such as Denial of Service attacks.
A single WAN link's availability is estimated to be no higher than 99% to 99.5%. To achieve the
desired IP Telephony availability, it is important to implement a strong network redundancy
strategy. This suggests the use of backup WAN links, preferably provided by different service
providers, along with network management techniques which assist with rapid detection of
network failures and identification of the failed component(s), and a fail over strategy to
enhance the availability of the WAN, potentially raising it up to 99.9 to 99.999%.
For detailed information on network design processes and methodologies for engineering highly
reliable data networks, please see Section 3. Other references for properly implementing Avaya
IP Telephony solutions include:
http://www1.avaya.com/services/whitepapers/planningdesign.html
http://www1.avaya.com/enterprise/news/docs/thought_leadership/best_practices.html.
In order to prevent long service interruptions as a result of WAN link outages or frequent link
flapping in a remote branch office, a Local Spare Processor (LSP) can maintain local services.

Example: a Geographically Distributed Solution

When projecting the Avaya IP Telephony solution's availability, both hardware failure and
software downtime contributions are considered. Then, the impact of network failure on the
potential overall downtime is estimated. The full system availability is assessed based on the
sum of the downtime minutes experienced due to a subsystems failure on the path from point A
to point B and the fraction of the traffic that traverses this path. The following examples examine
the contribution of such solution to the enhancement of full system availability.
7 For desired higher availability, enterprises should negotiate a Service Level Agreement (SLA) with their service
provider guaranteeing 99.9% availability per link. As a result, a fully redundant link can meet a better availability
number. For example, Sprint announced its new SLA structure availability of 99.9% at ITU Telecom World 2003
in Geneva (Network World 10/20/03, page 31, www.nwfusion.com)
Design for High Availability
7
Issue 3.4.1 June 2005 261