Resolving accelerator topic in v12

Hi,
We are moving from v11 to v12 this year. One topic in this project is getting rid of the accelerators. Does anyone has experience with it? Did you do it yourself by some "Best practice" case? Or who did help you with that?

Some insight information by side: The accelerators for v12 will be removed end of 2024 / beginning 2025.

Thanks for your input, in advance.
Sylvana

Find more posts tagged with

Infor EPM Discussions

Comments

martin-reischl

Hi Sylvana,

the basic approach is to check your rules for any performance issues that may arise by simply turning off the accelerators.

There is some valuable documentation available on optimizing the "prefetching" rules for performance:

https://docs.infor.com/depm/2024.x/en-us/depmolh_cloud/default.html?helpcontent=olap_rules_engine/uuh1510174534586.html

https://docs.infor.com/depm/2024.x/en-us/depmolh_cloud/default.html?helpcontent=olap_rules_engine/afk1510174500406.html

Optimizing rules often means splitting one big rule into several smaller ones, e.g. by using subsets to restrict the target area, and getting rid of "if" clauses or nested formulas.

Then make sure you test your reports and queries for any remaining performance issues. Performance really is the main issue here - we really see situations where the calculation results are different (mainly where we relied on missing accelerators for no rules calculation, which was a bad practice anyway).

Of course there's consultants who have been through this in many projects, and could help. Obviously, they'll take a part of the budget for it too.

Best Regards

Martin

Daniel Thiessen

Hi Martin,

Thank you for your comment and referring to Infor Consultants and their experience.

Do you observe a typical tipping point where a customer should work together with Infor (Partner) to solve the No-Accelerators challenge vs. working on the challenge on its own?

[quote userid="59324" url="~/infor-epm-user-group/f/infor-epm-discussions/34739/resolving-accelerator-topic-in-v12/91181"]we really see situations where the calculation results are different (mainly where we relied on missing accelerators for no rules calculation, which was a bad practice anyway)

Can you share how you compare the results before and after the change? Is there a systematic way to capture all differences, to not rely on random samples only?

Best Regards

Daniel

John Robson

Daniel

There is the ability to turn on Rules Debugging - see https://docs.infor.com/depm/2024.x/en-us/depmolh_cloud/default.html?helpcontent=olap_rules_engine/wss1599845527887.html

You should have a test report that triggers the calculations you want to test. then clear the value cache (which can be done by editing a rule and adding or deleting a dummy space at the end of an rule)

Activate logging

Run the report - assess subjectively if it's OK

Deactivate the log.

Look at the rule start and rule end events in the log. You are interested in the difference. Rules are numbered in the order they are defined - I think it's zero based and remember that rules can call rules.

In my experience a small number of rules are slow and it's rare that significant optimization isn't possible.

John

martin-reischl

Hi Daniel,

>Do you observe a typical tipping point where a customer should work together with Infor

I don't see a general rule here - if you feel overwhelmed by the task, it probably makes sense to bring in someone with more experience, instead of spending days with trial and error yourself. You may want to get some help as a starter, or get end-to-end assistance. It all depends on your preferences and level of experience.

>Is there a systematic way to capture all differences

In fact, there is no way to retrieve all rules calculated values from any real-life cube. So you will need some reports that you will compare, to focus on real-life use cases.

Regards

Martin

David Swinburne

There are more useful resources within the Educational Materials section of the EPM User Group Community.

dEPM-712 - Series - OLAP Rules Engine - Part 1 - Calculations in OLAP

dEPM-713 - Series - OLAP Rules Engine - Part 2 - Best Practices in Writing Rules and Additional Best Practices

dEPM-714 - Series - OLAP Rules Engine - Part 3 - Cell Calculation Information

dEPM-715 - Series - OLAP Rules Engine - Part 4 - Rule Debugging

The key challenge on disabling accelerators is the identification and refactoring of rules that are constraining performance; the most effective method of achieving this is through the use of the Rule Debug feature which provides detailed metrics on each OLAP rule as it is executed.

A typical approach would be to identify a representative set of application use cases (typically reports) and to then review the Rule Debug output that is generated when each of the use cases is run (against a cold Calculated Values Cache).

There are common patterns which are known to constrain performance and these are discussed in dEPM-715 in addition to a detailed explanation of the Rule Debug feature.

Regards

Dave

Daniel Thiessen

[quote userid="60114" url="~/infor-epm-user-group/f/infor-epm-discussions/34739/resolving-accelerator-topic-in-v12/91192"]A typical approach would be to identify a representative set of application use cases (typically reports) and to then review the Rule Debug output that is generated when each of the use cases is run (against a cold Calculated Values Cache).

Hello David,

Thank you for your comment and referring to the video resources. You were already positively mentioned with your rules' expertise as part of the Infor universe :-)

From what I understand as the recommended approach, to test a portion of all values in a cube (by focussing on the relevant reports) is not good enough for what we consider as 100% quality. We have the old SEGMENT cubes used for Financial Consolidation and in recent years published figures to banks and auditors, so we can't just update the rules procedure and assume that there is no change to the data as our main reports seem to work after the change.

Is there any customer that did a non-manual comparison of cube values before an after the rule changes? Did you use AppEngine to compare the values?

Regards

Daniel

dEPM-655.mp4

Nick Gladkov

I also wanted to quickly mention the other topic we discussed recently which is related to Sylvana's topic - community.infor.com/.../accelerators-former-rules-engine-are-not-supported-in-olap-modelling

David Swinburne

Hi Daniel,

The approach I typically use is to create a slice in Self-Service for the target area of the rule I am refactoring; I then export this to Excel and compare it to the results of the refactored rule using Excel formulas. If a rule refactor impacts a target area of a 1000 cells then I reconcile the 1000 cells to ensure they deliver identical results and accept that any derived values return the same results as the unchanged rule.

Regards

Dave

Copyright © 2025 Infor. All rights reserved.