Avaya Application Solutions Deployment Manual page 151

Table of Contents

Advertisement

Communication Manager uses a first-site-preference algorithm for outbound trunk calls. This
algorithm specifies that all outbound calls first attempt to seize a trunk within the originating
station's site, and tries to use a trunk in a different site if and only if it is blocked at its local
trunks. For inbound PSTN trunk calls, the CO selects the trunk. Therefore, Communication
Manager cannot use an analogous first-site-preference algorithm for inbound calls. However,
such an algorithm can be effectively imposed by assigning different calling numbers for the
three sites, which is typical in this example since the sites are in different area codes.
The goal of a first-site preference algorithm is to minimize intersite traffic. When this algorithm is
used, there is intersite traffic if and only if it overflows to a trunk on another site after having
been blocked at the trunks in its own site. Under the assumption that a first-site preference
algorithm is used in this example, the trunks at the three individual sites must be sized
independently, as opposed to all together. Initially, the overflow traffic is ignored, but that topic is
discussed later in this example.
Since overflow traffic is ignored for the time being, intersite trunk traffic is zero, which implies
that the off-diagonal entries of the inbound and outbound COI matrices will all be zero. To
determine the values of the diagonal entries, which correspond to intrasite trunk usage, the
Uniform Distribution model is applied. In particular, 65% (that is, 1755/2709) of the stations are
in Atlanta, 24% (that is, 648/2709) of the stations are in Boston, and 11% (that is, 306/2709) of
the stations are in Cleveland. Therefore, the Uniform Distribution model implies that 65% of the
75 Erlangs of inbound CUR (that is, 49 Erlangs) is assumed to terminate in Site 1 (Atlanta), 24%
(that is, 18 Erlangs) is assumed to terminate in Site 2 (Boston), and 11% (that is, 8 Erlangs) is
assumed to terminate in Site 3 (Cleveland). Similarly, 49 Erlangs of outbound CUR is assumed
to originate in Site 1, 18 Erlangs is assumed to originate in Site 2, and 8 Erlangs is assumed to
originate in Site 3.
It is instructive for this example to construct three different COI matrices rather than just one.
Specifically, it is useful to construct one for intercom CUR, one for inbound CUR, and one for
outbound CUR. The information from the previous paragraph can be used to populate the
following inbound and outbound COI matrices
Distribution model in Example 2: Uniform Distribution
Table 22: Inbound COI matrix for the Uniform Distribution model in
Uniform Distribution model
Inbound CUR
From trunks in Site
____
(Table 22: Inbound COI matrix for the Uniform
model):
To stations in Site __
1
49 Erlangs
1
0
2
0
3
Call usage rates
Example 2:
2
3
0
0
18 Erlangs
0
0
8 Erlangs
Issue 3.4.1 June 2005
151

Advertisement

Table of Contents
loading

Table of Contents