21.7
NETWORK MANAGEMENT SYSTEMS IN ENTERPRISE NETWORKS
553
completing through the network is very low and the outgoing trunk groups or connected
switches are overloaded, selective protective controls can prevent wasted usage of these
overloaded network resources for traffic to HTR points. SDOC responds to switching
congestion by dynamically controlling the amount and type offered to an overloaded or
failed switch. STR, conversely, responds to trunk congestion in the outgoing trunking
field and is triggered on a particular trunk group when less than a certain number of
circuits are idle in that group.
SDOC and STR are two-level control systems. The first level indicates less conges-
tion than the second level. The first-level response is typically limited to control of
traffic destined for HTR points, whereas the second level applies controls to both HTR
points and other traffic, typically alternate-routed traffic. HTR traffic can also be manually
enabled.
HTR traffic is automatically detected by the AT&T 4ESS switch based on an analysis
of destination-code completion statistics. This analysis is performed on a 3- and 6-digit
basis every five minutes. In Nortel DMS-100 and DMS-200/500 switches, HTR codes
can also be manually selected and enabled.
Automatic controls, such as SDOC and STR, are intended to be activated by a switching
system within a matter of seconds in response to a switch or trunk-group overload. These
controls provide rapid protection for the network and, by their code-selective basis, attempt
to restrict traffic that has a low probability of completion. When automatic controls trigger,
network traffic managers monitor their operations and adjust system parameters to deal
with the particular network condition, whether it is a general overload, a mass call-in,
a natural disaster, or a major network-component failure. Among these parameters are
call-completion determinations that designate a code "HTR" and control-response options
that designate the amount of traffic to be controlled or trunks to be reserved at each
triggering level. Since the optimum control response depends on the severity, geographical
distribution, and type of overload, maximizing the calls carried by the network requires
coordination and combination of automatic and manual control responses.
21.7
NETWORK MANAGEMENT SYSTEMS IN ENTERPRISE NETWORKS
21.7.1
What Are Network Management Systems?
Ostensibly a network management system provides an automated means of remotely
monitoring a network for:
z
Quantification of performance (e.g., BER, loss of synchronization, etc.)
z
Equipment, module, subassembly, card failures, circuit outages
z
Levels of traffic, network usage
The impetus of the systems described below has come from the enterprise network envi-
ronment, typically from the developer of TCP/IP (US DoD) and later from the OSI
world. There are now many proprietary network management systems available to the
user. Among these are Hewlett-Packard's Openview, IBM's Netview, and Digital Equip-
ment Corporation EMA (Enterprise Management Architecture). There has been a distinct
trend toward distributed processing in the network management arena.
Such network management systems, especially in the distributed processing environ-
ment, require a means to communicate for monitoring and control of the enterprise
network. Four network management protocols have evolved for this purpose, which we
describe below.