DP Live Cache question

[Former Member]

Hello DP experts

I am trying to understand role of live cache capacity in Demand Planning speed and performance - the factors of  number of CVCs, users , batch jobs etc.

Question1  - if a planning area is added whereby let's say CVCs , batch jobs and users are all doubled in an existing DP environment, is it fair to assume that the required live cache memory should also roughly double for the same level of performance?

Question 2 - All the planning area data is already in the live cache. What exactly happens structurally when a selection is loaded in the data view-does live cache capacity play a role in speedily loading the selection in the data view? or is it something like  just the CPU processing speed- outputting the data to the screen?

Thanks

Anton

[alok_jaiswal · Accepted solution]

Hi Anton,

1. Memory used in Live cache for DP is at high level dependent on Versions, planning horizon, periodicities/planning bucket profile, key figures and CVC in your system. Time series network is created in LC for each CVC combination and it depends on the horizon which was used during initialization. For better memory and performance management, it is recommended to  define horizon accordingly and also run report /SAPAPO/TS_PAREA_INITIALIZE with dynamic variant which helps in  periodically moving the initialization horizon of planning areas into future and deleting last period from Live cache by shifting the time series data.

2. During loading of data along with Live cache - data in planning area plays major role. For example how many CVCs are there in planning area - at what level of aggregation are you trying to load the data, are there many macros defined in your planning books which needs to be executed like default macros, how many key figures defined in the planning book etc. All these factors also play major role in how much time it takes to load data in planning book.

Hope it helps.

Regards,

Alok

[Former Member · Accepted solution]

Hi Alok

Thanks for the general concept and explanation.

My specific question is if planning area 1 has all the parameters you mentioned (version, CVC, Time series, macros etc) and users, and then area 2 is added with same or remarkably similar parameters and users, will live cache requirement nearly double? I know this is not a rocket science but roughly?

Regarding my second question - example again. If I have live cache memory X, and then loading a selection in the data view with a combination of 6 characteristics,resulting in say 200 products out of 800, and it takes 3 seconds. All 800 product data is already in the LC before selection is loaded. Now all things remaining the same, if LC memory is doubled to 2X will this loading be faster? I guess my disconnect may be between LC memory role and processing speed.

Regards

Anton

[alok_jaiswal · Accepted solution]

Hi Anton,

Yes for your first requirement if you are copying one more set of planning area and which in turn contains all the key figures, CVCs etc. for same time horizon then memory occupied in Live cache will also increase proportionally (approx.)

For your second requirement, I cant specifically say if doubling the LC memory will result in faster data load - obviously if LC is healthy then jobs and proceses tend to run faster. Also the speed observed by users while using various functions interactively depends on various underlying technical parameters too like Network etc. So as you said LC memory and processing speed may not be directly proportional but you can notice some improvement.

Hope it helps.

Regards,

Alok