2. USAGE CONSIDERATIONS › 2.1 Introduction to the Use of Network Data

2.1 Introduction to the Use of Network Data

Today, almost no large installation can function without
sophisticated data communications capabilities.  Those
requirements pervade the organization so completely that
inadequate performance of the communications facilities can
adversely affect the profitable operation of the business.
Because of this dependence, it has become mandatory for
organizations to make a conscious effort to manage their
communications networks systematically.

The management of large-scale communications networks is a
difficult task because communications is an integral part of
so many divergent activities.  The scope of resources,
vendors, applications, hardware types, tools, problems, and
users exceeds that of any other group in the data center.

This management effort is further hampered by the lack of an
established methodology (such as exists within the mainframe
environment) for performance evaluation, tuning, and capacity
planning.  Until recently, there existed a shortage of
comprehensive network measurement tools capable of producing
usable network management information.

IBM and other vendors have addressed this lack of data with a
series of network management products.  However, these
products lack integration, which can result in redundant
network data.

CA MICS addresses this problem by:

o Providing a common database to support the multitude
  of data sources

o Integrating network monitor data with application
  monitor data

o Supporting a network management methodology that is
  integrated with an overall installation management
  methodology


o Supplementing the existing monitor data with calculated
  and user-specified characterization data to increase
  its usability

o Producing a variety of tabular management reports which
  are not available from the network monitor products

o Providing analysis of network data using CA MICS exception
  monitoring and reporting.

The network monitor products produce three broad categories
of data:

o Network Load and Utilization Data

o Network Response Time Data

o Network Reliability Data

Network load and utilization data consists of command, path
information unit (PIU), and character counts by logical unit
(LU), physical unit (PU), link, application, and so on.  This
data can be summarized to indicate the amount of network load
that is supported or generated by a given network element.
Link workload may be expressed as a function of link
bandwidth (speed), thus yielding a percentage utilization
figure.  Percentage of link utilization reported over time
for key network links can be used to anticipate the
development of bottlenecks and estimate the magnitude of
required upgrades.

Network response time data consists of host, network, and
total response times by LU, PU, link, application, and
performance class.  These network service statistics can be
analyzed in light of installation standards and industry
accepted values to indicate the adequacy of end-user service.
Service degradation can be identified as residing in the
network or in the host processor.  In many cases, service
problems can be isolated to a given network application,
path, or component, thus limiting the scope of activity in
formulating a solution.

Network reliability data consists of error statistics,
events, and alerts resulting from hardware component
malfunctions.  Error statistics are made up of temporary
error counts and network traffic information.  Events are
recorded when the ratio of temporary errors to traffic
exceeds an installation-defined threshold or when a permanent
error occurs.  Alerts are recorded for installation-defined
error conditions of extreme concern.  Additional reliability
data is provided in the form of errors which result in
premature UNBIND conditions between primary and secondary
logical units.