Table of Contents
Advertisement
The following table shows the result of the calculation for duplex S8700 servers in active/
standby mode.
Table 54: Failure perceived by enterprise due to redundancy
in the S8700 Duplex Server Complex
Sub-System
Single S8700 Media Server
Redundant S8700 Media
Servers
1. Telecordia GR-512 Core: Equipment Service Restore Time assumption indicates a
In the case that the redundant component is implemented in active/active mode such as the
power supplies in an Avaya G650 Media Gateway, the Markov state-transition model changes
slightly. In this scenario, upon failure of one component, the parallel component is compelled to
carry twice the original load, which means that the corresponding failure rate is estimated to
double. This increased failure rate is captured by substituting 2 for the failure transition rate
from State 1 to State 0.
C. Individual Sub-System Availability and Annual Downtime Minutes
The component's Mean Time Between Failure (MTBF) data together with its Mean Time to
Repair (MTTR) is used to accurately project individual sub-system availability. The result will
lead to calculating the annual downtime minutes experienced due to the sub-system failure.
The calculated downtime is prorated according to the percentage of the end-users of the
sub-system, or more accurately, the percentage of the total traffic generated in that sub-system.
For example, consider a configuration with a remote media gateway in a branch office that
generates approximately 10% of the total enterprise traffic. Upon failure of the Avaya Media
Gateway serving the branch office site, the contributed downtime of that gateway being off-line
to total system downtime is estimated to be
th
i
SubsystemContributedDowntime
The prorating factor of 10% comes from assuming full service interruption of 10% of the full
system generated traffic or in the case of the Uniform Distribution Model assumption, 10% of all
the enterprise end-users. The prorating factor of 10% (90% comes from assuming the rest of
the enterprise end users, can not make calls, which terminate into the failed Media Gateway
component of the enterprise solution.
total of three (3) hours of service restore time for attended sites (those with onsite
technical resources), and four (4) hours of service restore time for unattended
sites (those sites to which technical resources must be summoned). In many
cases the hardware failure discovery and repair time can be shorter due to the
alarming and error logging capabilities implemented on all Avaya products.
Mean Time Between Failure (MTBF) in
Hours Assuming a 4 Hour Mean Time To
1
Repair
52,815
8
3.487
10
==> Availability = 99.9999+
=
RemoteGatewayDowntime
Design for High Availability
10%
+
10% 90%
Issue 3.4.1 June 2005
259
Advertisement
Table of Contents
