Monitoring Network Performance; Routing Probe Traffic Using Policy-Based Routing (Pbr) - HP 7102dl - ProCurve Secure Router Configuration Manual

Network Monitoring
Overview
The ProCurve Secure Router allows several types of probes to test routes. You
can use ICMP echo probes to test simple connectivity to a remote device. Or,
you can use TCP connect or HTTP probes to test connectivity to a particular
application on a remote server.

Monitoring Network Performance

Network monitoring on the ProCurve Secure Router is designed primarily to
test connectivity. However, you can use network monitoring to create limited
performance tests. For example, you can set the timeout value low in order
to test the link speed. When the link is too congested, the probe returns a
failure to the track.
Sometimes you will want a backup connection to become active, and some-
times you will simply want to monitor the congestion. In the second case, you
should not associate the track with a route. Simply enable the track to create
a log each time the probe indicates unacceptable performance. You can enable
the ProCurve Secure Router to save the log to event history and to forward it
to an external server. (See "Enabling a Track to Log Changes" on page 9-29.)
TCP connect and HTTP request probes can perform limited tests on the status
of a particular network resource. For example, you could configure the probe
to fail if a remote server does not return a response within a set amount of
time, indicating that the server is overburdened. For Web servers, you can
configure the HTTP probe to fail when the server does not return the correct
status. Again, the track can either remove a route to make way for a secondary
route, or it can simply log the failure so that you can take corrective action as
needed.

Routing Probe Traffic using Policy-Based Routing (PBR)

When you use tracks to monitor routes, you must use PBR to route probe
traffic. Creating a route map for probe packets ensures that the probe moni-
tors the connection that you want it to monitor at all times.
When a track that is monitoring a route fails, it removes that route from the
routing table. However, the probes remain active. This behavior allows the
track to reinstate a route after a failure is remedied; however, it can also cause
a problem: reinstating the route before it is actually fixed.
The track has removed the failed primary route from the routing table, so the
backup route is added to the table, and traffic can again reach its destination.
Because probe packets can also reach their destination over the backup
9-9