Monitoring Domino Application Servers
eG Enterprise has developed an exclusive Domino application monitoring model (see Figure 1) for the Domino application server, which employs an eG agent to continuously monitor the performance of the server. This eG agent executes tests which communicate with the Domino SNMP agent to collect key metrics relating to server performance. To facilitate this communication, SNMP should be enabled on the application server.
Once the Domino SNMP agent is completely configured, the tests executed by the eG agent (which is monitoring the Domino server's performance), polls the SNMP agent at frequent intervals for performance data. The SNMP agent then retrieves the desired performance statistics from the SNMP MIB of the Domino server and forwards the same to the eG agent.
Using these statistics, administrators can easily and accurately answer the following performance queries:
- Has the Domino server's capacity been fully utilized? If not, what percentage of its capacity is still available for request processing?
- Are sufficient threads available for handling requests to the Domino server?
- Are there too many concurrent connection requests to the server?
- Has enough memory been allocated to the Domino processes and shared memory segments?
- Are the NSF buffer pools and buffer control pools adequately sized on the Domino database?
- Are hits to the database cache optimal, or does the cache require any resizing?
- How many databases are currently awaiting replication and for how long have they been waiting? Were too many databases in the queue for too long? If so, can additional replicators be configured to share the load?
- Has the cluster replicator successfully handled all replication requests to it, or have too many requests failed?
- What is the current load on the web server component of the application server? Is it too heavy?
- Is the idle session count kept at a minimum?
The tests that reveal the above are mapped to the various layers of the monitoring model of Figure 1.
The section that follows discusses the first layer of Figure 1 and the tests that are mapped to it. While the next 2 layers have been dealt with extensively in the Monitoring Mail Servers document, the remaining layers have been elaborately discussed in the Monitoring Unix and Windows Servers document.