TM1 Forum

Hi,

I just a few questions about TI process locking. I am asking this because I want to build a number of user processes that users can run on the web and want to know how to avoid any locking. Please assume that Parallel Interaction is enabled.

Example A (no meta data updates):

I create a process that puts values to a cube and the process creates views on the fly using hard coded view and subset names.
User A runs the process and immediately after User B runs the process. Will User B be locked out until User A's process has completed?

Example B (no meta data updates):

I create a process that puts values to a cube and the process creates views on the fly using dynamic view and subset names, i.e. includes the user name in the view and subset name.
User A runs the process and immediately after User B runs the process. Will User B be locked out until User A's process has completed?

I would understand in Example A User B being locked out and would expect in Example B that User B will not be locked out because the View and Subset names are not identical. Are my assumptions correct?

Maren

Hi Maren

When creating the views and subsets on the fly, try creating temporary subsets and views. It doesn't matter what they are called since they only exist within the user's session and for the duration of the process. We found that this gave a dramatic reduction in locking. If you create permanent subsets and views then that is a meta data operation and it will cause locking.

Regards

Paul Simon

What is the purpose of the views and subsets ? Used in the process to copy data and so a zero out, or views that will be used later in by the user ? If the latter, views and subsets need to be permanent. If the former, indeed, you should use temporary objects.