The last two monthly tips talked about APM CE configurations that have troublesome implications. This month, we look at the success rate trap.
First let us review what the APM Glossary defines as a success rate.
The success rate measures the percentage of transactions that are defect-free. CA CEM Calculation: success rate = 1 - defective count / total count
Note the key words "percentage of transactions that are defect free." This is talking about ALL transactions and ALL types of defects.
Not realizing this, the unwary APM administrator gets in trouble. They agree to comply to a very high SLA with their application owners thinking that this applies to slow time only
But if the network data is not cleaned up, they could get a high number of missing and partial transactions. Or see a high number of Client or Server errors due to network and server issues.
All this contributes to a low success rate even though there may be a low number of slow time defects.
Discovering too late their plight, the frantic APM Admin opens up a support case asking what to do.
Currently there are three things that can be done:
1) Ask for an enhancement request in this community site for a slow time success rate be added to a future release.
2) Create a custom report that showing the slow time success rate for a business transaction.
3) Use Yield as a measurement instead which is less impacted by this situation due to the larger denominator than a success rate.
Questions for Discussion:
1) Do you know an APM Administrator that fell into this trap?
2) What did they do about it?
3) What other troubleshooting topics would you like to be covered in Tuesday Tips?