TI processes significantly slower after upgrade from TM1 10.2.2 to TM1 2.0

[AskAndAnswer]

Hello,

I have a bunch of processes that significantly slowed down after an upgrade from 10.2.2 to 2.0. The process that used to take 2 minutes now takes 20. There are no changes in the configuration file, and ForceReevaluationOfFeedersForFedCellsOnDataChange is set to "T" on both old and new servers. Not all processes are affected, but mostly those that use TM1RunTI to trigger a process in another database.

The new server is a much bigger machine, so there should not be any RAM issues.

What could be a reason?

Thank you.

[jim wood]

What in general are they doing? There may be a common thread.

[declanr]

When you say the processes that are generally slower are those that are triggering work on another Database, do you mean another tm1 service or something on a relational DB?

If it’s on another tm1 service, is that on the same Server and if so is every process on that service slower?

If it’s relating to a relational DB, then it’s worth looking at the speed of the network between that and your TM1 server.

[AskAndAnswer]

It is another TM1 database that's running on the same server. The processes allocate data that's loaded from files that are also on the same server.

[declanr]

Start by narrowing down which part of the TI is running slower, easiest quick way to do this is to put in asciioutputs with time stamps at the end of main sections in the process.
There will be plenty of things that could make it slower, the more you can narrow it down the more suggestions you will get (or at least better quality suggestions.)

If you have multiple TIs running slower try and do this in all of them and see if it’s a common bit of code. From what you’ve said it sounds like you have TIs doing an output to a text file on one service and the other service then does the import. Perhaps it’s slower writing to the file or perhaps it’s slower reading the file in.
It’s unlikely (as it would probably error rather than be slow) but maybe IF where it’s slow is writing to the file, it could be something like a virus scanner being rather overzealous.

Also watch ops console while the processes are running, maybe there is a lock contention occurring and the processes are rolling back a few times before they actually succeed.