I wanted to understand the Inter Tier Time of the following screenshot.
One transaction has taken 45.68s and the App tier contribution is 50.61%. On the other hand, I can see that the inter tier time for 10x calls to another tier (middleware) is 40.45s and the total contribution is 44.82%.
Is this due to the network between both tiers or something else?
Inter Tier Time/Transaction: Average latency time for transactions in milliseconds and total transaction execution time percentage.
InterTier time is difference between last call on first component and first recorded call for the same transaction recorded on second component. So In general it can be all, network, firewalls etc. I've had problems with InterTier time for enviornments that was working on single machine and were communication via localhost, there were problems with disk performance, in general app was slowing down during such communication and inter tier time was high as well.
Hello @sebastian k.
It means we can consider that the inter time was spent due to a network or firewall (if exist) in between both tiers. The communication is between IBM WebSphere Application and IBM Integration Bus (WebSphere Message Broker).
What else can be considered in this situation?
Like I've said, appmon compares time of leaving 1st component and reaching 2nd component. So everything that can slowdown transactions. I've mentioned example for single host application where disk was root cause 🙂 for architecture that is on multiple hosts I don't have extra idea right now.
In general - bigger resource contention is, higher intertier time you have.
I.e. it's a question of balance between # of workers / java (.net) threads / vCPUs and bare metal CPUs at the end. I mean in most of the cases it's the application (and its infra) problem.
Your case is about WSMB. An agent there, just "instruments" on the EXIT_PATHs. It means you have limited visibility of process definition and its end-to-end performance.
Scratch an extra square centimetre, to see what is the incoming request handler there 😉
In case of internal Data Centers and traffic encapsulated there, mostly never I see this problem because of NW active infrastructure.