MAPI revisited

Back on this again (been a few days)....

I was successful in configuring a template for a MAPI test but was dismayed by the unresonable response time (over 16 secs).

I decided to look into this further but came across an another issue which has me perplexed. When I went back into the template (you will see it is green) and select a test node (the same one) it errors out (see error).

What am I missing or not understanding? Why won't the test work when it is already running?

Also, I know that it isn't taking 16 seconds for our mail to flow but how can I troubleshoot the MAPI above to get a true representation of response time?

Find more posts tagged with

response_time

mapi

Accepted answers

All comments

FormerMember

You can run a test while the application monitor is running. The fact that your failed is not a systemic problem; it may have failed for some transitive reason. If it never works for you, please open a Support ticket, because it should.

Why 16 seconds? That time includes everything we're doing. It includes the time it takes us to the connect to the SMTP server, send the email, start up a secondary MAPI "helper process", have the helper process open a MAPI connection and check for the email in the receiver inbox. The time represents all of the effort, not just the actual A-to-B travel time for the email. It is a "true" representation of what it takes to send an email. Aren't you more interested in increases from this baseline time than the absolute time?

FormerMember

Thanks Denny for the reply. As for "Aren't you more interested in increases from this baseline time than the absolute time?" - No. This number is unrealistic. If it took 16 seconds for our email to flow then our users would be screaming!

Unfortunately I am comparing this to what we were using which was HP's OVIS product which uses probe agents to send a test email to the Exchange server and measure the time it takes to reply back. These numbers usually measure less than 2-3 seconds. We based contractual agreements based upon email transactions taking less than 6.2 seconds - measured by tools like OVIS. We were hoping to replace OVIS with APM but this looks like this is not possible. We would really like to get APM working to do this but doesn't look promising....

FormerMember

As I said before, we're measuring more than the flow time. Sending an email is also more than just the flow time. HP is looking at a different measure, so you're comparing apples and oranges. I understand that if you've based contracts on the measure that HP is giving you, it's difficult to switch.

josh.clark

I was noticing on some of my test machines that the mapi client can affect the test speed. On machine with Outlook installed, the mapi test was faster than on machines with the free client.

Can you try installing Outlook on the Orion server? I'm not sure if it will improve the monitor's speed or not, but it seems like a fairly simple test.