Page 383 - Handbook of Modern Telecommunications

3-174 CRC Handbook of Modern Telecommunications, Second Edition

3.7.2.2.2 Core Management Services
The management framework is expected to offer core services and interfaces to other applications. The
basic management applications to be provided are discovery/mapping, alarm management, and plat-
form protection.

3.7.2.2.2.1 Discovery and Mapping Device discovery/network mapping discovery refers to the network
management system’s ability to automatically learn the identities and types of devices currently active
on the network. At minimum, a management platform should be capable of discovering active IP devices
by retrieving data from a router’s IP tables and Address Resolution Protocol (ARP) tables.
However, even this capability does not guarantee that all IP devices on a given network will be
detected. For example, relying solely on routing tables is inadequate in purely bridged networks where
there are no routers. Thus, a more comprehensive discovery facility should also include other mecha-
nisms such as broadcast messages (ping and others) that can reach out to any IP device and retrieve its
address and other identifying information.
On the other hand, discovery mechanisms that rely completely on broadcasting (e.g., ping) will incur
a tremendous amount of overhead related to the process of locating devices on the network. Ideally, a
management platform should support a combination of ARP data retrieval and broadcasting.
A complete network discovery facility should be capable of detecting legacy system nodes as well.
At present, most platforms rely on third-party applications or traffic monitoring applications to supply
discovery data on non-TCP/IP devices.
Another desirable feature is the ability to run automatic or scheduled dynamic discovery processes
after the initial discovery, with the goal of discerning any changes made to the network after the initial
discovery has taken place. In large networks especially, overhead and consumed bandwidth for running
a dynamic discovery process continually in background mode may be too great; therefore, the ability to
schedule discovery at off-peak hours is important.
It is also important for the user to have the ability to set limits on the initial network discovery. Many
corporate networks are now linked to the Internet, and without predefined limits a discovery applica-
tion may cross corporate boundaries and begin discovering everything on the global Internet. Some
management platforms allow users to run discovery on a segment-by-segment basis. This can prevent
the discovery process from getting out of hand too fast.
Many management platforms are capable of automatically producing a topological map from the data
collected during device discovery. However, these automatically generated maps rarely result in a useful
graphical representation. When there are hundreds of devices, the resulting map can be very cluttered
in appearance and be of little use.
Even when the discovery process operates on a limited or segment-by-segment basis, there will eventually
come a time when the operator must edit the automatically generated network map to create a visual picture
that is easier for human beings to relate to. Therefore, the ability to group objects on the map, and move
them around in groups or perform other types of collective actions, can be a real time-saving feature.

3.7.2.2.2.2 Alarm Capabilities Management platforms act as a clearinghouse for critical status mes-
sages obtained from various devices and applications across the network. Messages arrive in the form of
SNMP traps, alerts, or event reports when polling results indicate that thresholds have been exceeded.
The management platform supports setting of thresholds on any SNMP MIB variable. Typically,
management platforms poll for device status by sending SNMP requests to devices with SNMP agents
or Internet Control Message Protocol (ICMP) echo requests (pings) to any TCP/IP device.
The process of setting thresholds may be supported by third-party applications or by the management
platform. Some platforms allow operators to configure polls on classes of devices; most require opera-
tors to configure a poll for each device individually.
A majority of platforms support some degree of alarm filtering. Rudimentary filtering allows opera-
tors to assign classifications to individual alarms, such as informational, warning, or critical, triggered