ISPPlanet
Network
Management System Series- SNMP
ToolChest
Status Monitoring
SNMP managers typically use repeatedly polling with SNMP GETs, complemented
by agent-generated TRAP or INFORM messages, to monitor network status.
Although it lacks a utility to receive TRAPs, the SNMP ToolChest makes
extensive use of poll-based monitoring and analysis.
SmartBrowser (right) can be configured
to poll several device interfaces and apply a single analysis expression.
Numerous pre-packaged expressions provide on-the-fly analysis of polled
MIB objects. For example, Ethernet, WAN, ATM, and Token-Ring expressions
include transmit and receive utilization, packet rate, and link errors
appropriate for each media. Expressions available at the IP level
include packet rate, ping rate, and the percentage of management traffic
and errors. A configurable threshold is compared to each expression
resultvalues that exceed this threshold are flagged in red.
Additional real-time monitoring tasks are
performed by RouterScope (left). Like SmartBrowser, RouterScope
accepts a list of IP addresses and polls these devices at a configurable
interval. But Router-Scope polls many routing-related objects at once
and performs more extensive analysis, computing forwarding rate and
comparing ICMP, IP, and Cisco-specific objects to hard-coded thresholds
(customizable by updating the Tcl).
Threshold violations are flagged and color-coded buttons provide additional
information about affected MIB objects. Using RouterScope, admins can spot
resource exhaustion, network congestion, and error rate spikes that may
indicate routing problems.
A third monitoring tool, IPScope (right), continuously
monitors end-to-end reachability. This unique tool actually plots
the forward and reverse paths between a pair of devices, then polls
each node and link in these paths. IPScope analyzes retrieved values
and compares them against GUI-settable thresholds: for example, %
CPU busy for nodes or % receive or transmit utilization for links.
Violations are indicated by icon color changes.
Display any device problem list by clicking on the affected icon, or use
the icon's pop-up menu to view node or receive/transmit stats. Using IPScope,
admins can spot end-to-end reachability problems and isolate the point of
failure, but only if every device in the path is SNMP-accessible. We ran
into one unresolved problem with IPScope: our Cisco routers returned a zero
next hop ifIndex that impacted IPScope's plotting algorithm. SNMP ToolChest
is still investigating.
These three tools provide significant added value over raw object polling
by performing real-time analysis and thresholding. But visual cues are
only effective when a human is watching; simple audible or email notifications
would make these tools even more powerful.