The Network SLAPM2 subagent provides information about service
policies and performance data for applications mapped to those policies
through the slapm2PolicyRuleStatsTable.
Note: The Network SLAPM2 subagent can be used to monitor Differentiated
Services policies.
- slapm2PolicyRuleStatsTable
- Provides statistics on a per policy rule basis.
The Network SLAPM2 subagent also supports performance monitoring
with the slapm2PRMonTable object. Entries are created in the monitor
table to establish the wanted criteria for monitoring. The following
level of monitoring is provided:
- Aggregate
- Monitoring is performed based on the aggregate of all TCP or UDP
applications that are mapped to one or more service policies.
Three types of monitoring are provided for measuring application
performance:
- TCP round-trip time
- The current TCP round-trip time of applications are compared to
the threshold values established in the monitor table entry. If the
current rates exceed the high threshold or go below the low threshold,
an SNMP trap is sent if traps are enabled.
- TCP packets retransmit ratio
- The current TCP packets retransmit ratios of applications are
compared to the threshold values established in the monitor table
entry. If the current rates exceed the high threshold or go below
the low threshold, an SNMP trap is sent if traps are enabled.
- Average accept queue delay
- The current average accept queue delay of applications are compared
to threshold values established in the monitor table entry. If the
current rates exceed the high threshold or go below the low threshold,
an SNMP trap is sent if traps are enabled.
For more details about how to make the various monitoring calculations,
see the NETWORK-SLAPM2-MIB in the sample file slapm2.mi2 in the /usr/lpp/tcpip/samples
directory.
When SNMP traps are enabled, and a not achieved trap
is sent as described above, a corresponding okay trap
is sent when the traffic again conforms to the boundaries established
in the monitor table entry.
For example, suppose the slapm2PRMonTcpRttDelayHigh value is set
to 2 seconds and the slapm2PRMonTcpRttDelayLow value is set to 1 second.
If the TCP round-trip delay rises above 2 seconds, a not achieved trap
is sent. If the TCP round-trip delay then drops below 1 second, an okay trap
is sent to indicate the problem has been resolved. However, if the
row becomes inactive before conforming to the established boundaries,
an okay trap is never sent, because this removes
monitoring for this entry.
The following traps are used to monitor table administration:
- Policy deleted
- A slapm2PolicyRuleDeleted trap is sent when an entry is deleted
from the slapm2PolicyRuleStatsTable, if slapm2PolicyDeletedTrapEnabled
is enabled(1).
- Monitor deleted
- A slapm2PolicyRuleMonDeleted trap is sent when a slapm2PRMonEntry
is deleted, if the value of slapm2PolicyDeletedTrapEnabled is enabled(1).