OSI/MHS Configuration and Management Manual

OSI/MHS Configuration and Management Manual—424827-003
8-1
8
Sizing and Tuning Your OSI/MHS
Subsystem
This section describes performance tuning for the OSI/MHS subsystem. Specifically, it
covers the following topics:
•
Symptoms of performance problems
•
Performance tuning steps
•
Guidelines for handling large messages
Symptoms of Performance Problems
The most fundamental symptom of a performance problem is the perception, by users
of the messaging system, that messages are not delivered quickly enough or that too
many messages go undelivered because they could not be delivered on time. In a
store-and-forward system like an X.400 network, response time is not usually a
prominent issue. However, users have expectations for how quickly a message should
be processed and a reply or acknowledgment returned.
Although prediction of performance problems is not an easy task, certain symptoms in
the operation of MHS can serve as indicators:
•
Reports of resource shortages
•
Repeated routing attempts
•
Large numbers of retries
•
Connection rejections
•
Consistently large internal queues
Reports of Resource Shortages
OSI/MHS reports an event (ZMHS-EVT-CROSS-THRESHOLD) when the use of
certain types of resources, such as memory pools or disk space, exceeds a certain
threshold. This event, if it occurs repeatedly, suggests that the sizing of the resource is
insufficient to provide adequate performance.
Repeated Routing Attempts
OSI/MHS reports an event (ZMHS-EVT-ACCT-MRP-ROUTE) whenever it attempts to
route a message, either to another MTA or to some local entity such as a gateway, the
message store, or a distribution list. Repeated routing attempts for the same message
can have several possible implications, such as unreliable connections, inefficient O/R
naming, or an overloaded adjacent MTA, all of which affect performance.