8. MANAGEMENT SUPPORT APPLICATION INTERFACES › 8.1 CA MICS Accounting and Chargeback Interface › 8.1.1 Chargeback Considerations

8.1.1 Chargeback Considerations

The CA MICS Accounting and Chargeback Option supports the
following CA MICS Network Analyzer files for chargeback
processing:

o  NCP Network Accounting (SNTNAC) File
o  CA NetSpy Terminal (SNTNSS) File
o  NPM User Activity (SNTPSU) File
o  NLDM Session Accounting (NVSNSA) File
o  NLDM Session Connectivity (NVSNSC) File
o  NLDM Response Time (NVSRTM) File

o  NLDM Route (NVSRTE) File

The NCP Network Accounting File can be built from either CA
NetSpy network session accounting data/network gateway
accounting data or Tivoli NetView Performance Monitor (NPM)
network session accounting data/network gateway accounting
data.  The CA NetSpy Terminal File is built from CA NetSpy's
VTAM interface data.  The NPM User Activity File is built
from Tivoli's NPM Session data.  The NLDM files are built
using data from either Tivoli's NetView Session Monitor/NLDM
or CA NetMaster Network Tracking System (NTS). The Telnet
file is built from CA NetSpy TCP/IP interface.

Using these data sources for chargeback extends the
capabilities of CA MICS by allowing the cost of the data
communications network to be recovered directly rather than
being absorbed as an overhead.  This section discusses the
advantages of using this approach.

From an accounting standpoint, it is always preferable to
recover costs based upon direct measurements rather than
allocating overhead charges because the factors used to
allocate overhead charges are estimates.  Direct measurements
eliminate guesswork and have the additional benefit of
varying immediately when system or environmental changes
cause the use of a resource to vary.

Although valuable, you must apply direct measurements only to
those resources that are directly measured.  In the case of
network data, only use of the network transport resource is
measured.  Significant errors in resource allocation would
result from using this data to apportion, for instance, CPU
consumption.  For example, you could assume that the
utilization of all data processing resources is proportional
to the amount of data transmitted on the network.  Experience
would show that many tasks, such as DB2 inquiries, could
consume vast amounts of CPU resources while generating
relatively little network traffic, either for the inquiry or
the response.

Limiting the use of network data only to the allocation of
the costs of the data transmission resources still produces
considerable latitude in the types of algorithms implemented.
This is because the algorithm must reflect the organization's
structural philosophy, which may or may not be reflected in
the design of the network.  The remainder of this section
discusses three chargeback scenarios for networks with
identical topologies that are operated by organizations with
differing structural philosophies.


THE EXAMPLE NETWORK

In the following examples, the network consists of two sites,
as shown in Figure 8-1.


                    T                            T
   --------        /                            /
   |      |   ----/                        ----/
   | HOST |---|  |-------------------------|  | ------- T
   |      |   ----\                        ----\
   --------        \                            \
                    T                            T

     SITE A                                SITE B


  Figure 8-1.  Example Network


Site A has the host processing resources and a communications
processor (37xx equivalent) while Site B has only a
communications processor.  Both sites have terminals attached
to the network.


DATA PROCESSING AS A CENTRALIZED CORPORATE SERVICE

In this case, data processing is viewed by the organization
as a tightly controlled, centralized service.  Decisions
concerning the type of service, level of service, network
topology, line speeds, terminals utilized, and so on are
made by the central communications staff.

The cost of communications is allocated to end-users based
solely upon the volume of data transmitted.  The total
communications costs are divided by the total traffic load to
produce a cost-per-byte factor.  This factor is then applied
to all traffic.  This factor is typically adjusted on an
annual basis.

Because the user makes no decisions concerning the line
speed, routing of transmissions, or service level of the
applications, these factors are not taken into account when
setting rates.  It is assumed that the design of the network
will maximize the cost/performance ratio as defined by the
corporate planners.


DATA TRANSMISSION AS A CENTRALIZED RESOURCE

In this case, data processing is viewed as a resource to be
shared equally by all entities within the organization.  The
corporate communications network provides the connection to
this shared resource.  Typically, the user is responsible for
the costs of the boundary circuits, terminal control units,
modems, and local support personnel.

Costs of the corporate network are allocated based on traffic
volume and distance from a virtual host processor resource
that is located equidistant from all users (despite the fact
that the real data center is located at Site A).  Charges can
also be based on a standard corporate network link speed and
can be adjusted for varying link speeds.


GUEST USERS

In this case, Site B is a paying guest of Site A.  Site B,
which can be an acquired subsidiary of Site A, gains some
advantage from using Site A's processing resources.

Site B bears the total cost of the data transmission to Site
A regardless of the distance.  Site A does not subsidize Site
B's communications costs, as in the previous example, even
though Site A can choose, install, and support all of the
communication equipment.

Costs are allocated by simply totaling the costs of
communication and dividing them by Site B's traffic volume to
arrive at a cost-per-byte factor, as in the first example.


SUMMARY

In each of these examples, only the actual communications
costs are recovered (for example, modems, terminals, control
units, links, support personnel, test equipment, and so on).
Host processor costs are still recovered based on an
application's consumption of host resources.

The differences in algorithms result from the interpretation
of what portion of the resources belongs to the corporation
as a whole and what portion belongs to the end-user.