Mitel MiVOICE BUSINESS, Engineering Manualline

Page: 134 / 381

Engineering Guidelines
120
When a system is sized such that the number of trunks is less than 1.5 times the number of
agents, the overall call rate will typically be less than 2.5 times the incoming (trunk) call rate.
When the number of trunks into the system is more than 1.5 times the number of active agents,
then the overall call rate rapidly climbs due to the multiple handling of the calls into and out of
the various queues. When an agent appears in several groups, as soon as he answers a call
in one group he must be made unavailable in all of the other groups. Similarly, when he becomes
free this must be applied to all of the groups to which he belongs. This adds significantly to the
processor load, and reduces the capacity of the system. When calls overflow on a path to
additional groups, a similar increase in processing occurs because calls have to ring in multiple
locations and then be removed when answered. The System Engineering Tool is designed to
handle systems with all of these features in use, but it is strongly recommended that System
Engineering or Professional Services should be contacted to determine the suitability of an
installation.
PERFORMANCE WITH RING GROUPS
The method of searching ring groups can have a significant effect on the overall performance
of a system.
Terminal ring groups are good for a small number of members, but can be extremely slow and
CPU-intensive with a larger number. Large ring groups should be configured for Circular hunting
for optimum performance.
Ring All ring groups can also have a large impact on performance. As every member of a ring
group requires a call setup in order to ring, and even though only one may be answered, each
of the call setups still has to be torn down, so the number of apparent calls in the system is
multiplied by the number of members in the ring group. Large ring groups should be configured
for Cascade hunting for optimum performance. Clustered ring groups can also help to improve
the performance of Ring All ring groups as processing is distributed across multiple controllers.
Clustered ring groups can also have impact on resources. When members are not local to the
ring group, the additional call setup is now an internal trunking call rather than a local call and
therefore consumes internal trunking resources. Ring groups should be configured in such a
way that the majority of members are co-located with the ring group to optimize performance
and resources.
For further information on ring group configuration to optimize performance and resources, see
the System Engineering Tool and the
System Administration Tool Help for MiVoice Business .
PERFORMANCE WITH HUNT GROUPS
The method of searching hunt groups can have a significant effect on the overall performance
of a system. Terminal hunt groups are good for a small number of ports (e.g. RADs) but can
be extremely slow and CPU intensive with a large number of members. Large hunt groups
should be configured for circular hunting for optimum performance. Selection of hunt group
type—for example, Voice, VoiceMail, RAD, etc.—also has performance implications, especially
with respect to auto attendant and IVR operation. Hunt groups containing auto attendant or
IVR ports should be configured as VoiceMail type groups to take advantage of automatic
camp-on; otherwise, in a high traffic site, you may experience system slowdowns if calls are
rejected by call control due to excessive processing.