The View, Response Time Statistics are useful.
Can these be logged to determine the overall performance of the system?
Are they stored in a table?
Any version of CA SDM.
The response time statistics feature is useful. It enables you to say exactly how long a particular action is taking for a specific end user.
It aims to answer two questions:
1) How long does this web client action take to complete?
2) How long did it take for this specific user and this specific time?
So it is useful to use for those specific scenarios. But not as useful for situations such as:
* How is the system performing in general right now?
* How are many users being impacted right now?
* What is the trend across the system?
However, you do have some options to get to this information.
1) Enable the "Key Performance Indicators" against "System" objects.
- - -
This example reports how long requests for updates, inserts, and deletes to BOP Virtual Database took to complete:
- - -
Note that this is from the system perspective. It tells you how well the application is interacting with the virtual database - which itself depends on the network and physical database. But in short, a poor performance here will likely have an impact on end users. But end users can get poor performance from other factors.
Still, it is a good place to start in order to look at the performance of the system.
2) Examination of the SDM /logs/stdlogs.
The stdlogs contain a wealth of information, albeit in a harder to read format.
In particular, monitor for the key phrase "milliseconds". As any query that takes longer than 2 seconds to execute will write this message into the stdlog.
A few of these are natural on a busy system, where people can enter almost any SQL query that they wish. However, if the number spikes above a "baseline" level then it may indicate a need to investigate.
There are tools which can monitor logs for specific phrases in real time, such as Bare Tail.
Tied into this, would be reports such as "pdm_vdbinfo" and "db_report" which give information about the state of agents and queries.
3) Moving into the more technical layers, you could enable products such as Wireshark or Fiddler. However, these are more administrator intensive and are usually only used in short, specific periods to monitor and understand a specific performance issue. They are not generally left running, as without constant interpretation, they don't produce meaningful data.
4) Finally, if you would like the "deluxe" option which is closest to answering your original query, then it would be to use a tool such as CA Application Performance Management (CA APM).
This DOES give you the ability to monitor the state of the CA Service Desk Manager in the level of detail that you asked for, which is monitoring how the end performance of the web clients is actually going, in real time, with alerts.
It may be more than you need, but here are some links so that you can understand what is available:
Try enabling (1) as the first step.
This will give you monitoring and reporting on the system performance that is useful as a starting point.