ALM - Best Practices
Hi Guys,
We have just started going though the documentation on Amapedia and make some testing for our organisation, as we are preparing to bring in ALM (we already have a production system). Those are some initial questions that poped up, before going even deeper in updating any referenced formulas. Would appreciate if yo can tell from your experience which would be the best practice for the below.
1. Import model vs Create from revision - assuming we already have the production model, can we make the Initial dev through an import instead? are there any inconveniences when synchronizing further on? I'd choose this as we want to have a 1 to 1 copy with the production, meaning also Production data - I would not achieve it through create from revision tag.
2. How long does usually take for a synchronization (60 -70gb model) - our first test took around 20-25 minutes for just a line item added in one list (is it normal or might be due the fact highlighted above?)
3. How do you bring in Data and MD in your Dev model? We do have daily integration with other systems that populates master data lists and also we would probably need to update Actuals every month - This creates problem if we want to keep 1 to 1 relationship with our dev model, as only production will e updated. Is it a common practice to make regularly copies of the production that will become a new Dev model? Is there an alternative?
Answers
-
Great questions!
1. Not at all. The first DEV instance can be a copy or import
2. In my experience, model size is not related to sync times. Best practices suggest moving small changes regularly to reduce complexity and time taken
3. You will have to create a separate set of integrations as well as schedules to keep DEV on par with PROD.
Regards,
Anirudh
0 -
Thanks for your feedback and insights.
For the sync part, we only moved a small change (one member in a list), that would normally take 2-3 minutes to make it in production. Doing the sync took around 20-25. Wondering the time would take for much more changes.
As for the integration, that would suit the MD loads, but not actual data inputs in Actuals or forecast, that are done solely in Anaplan Production model.0 -
I have a slightly different answers for you.
1. Dev model has to be as lean & small as possible making me inclined towards Create Model from Revision. Having Production Data in your Dev model is not recommended as you might expose your live data to people who you don't want that to see.
2. It depends. If you are pushing multiple Revision tags at once it might take a while but nonetheless it should not take that much time
3. I would keep Dev out of Integration and do that in my Test Env but if you don't have one you can probably test that in Dev and later on mimick it in Prod by changing the WS ID and Model ID in the scripts
Hope that helps
Misbah
1 -
I have noticed ALM syncs taking a long time when they add a significant amount of space to the model, otherwise adding a single line item certainly shouldn't take 20-25 minutes...
As to copying over user input data, regular model copies are the best way to do it.
However, there is another approach (which I recommend only as a last resort) to copy over user data inputs from PROD to DEV by making use of the Import Data Sources
1. In DEV, create import actions for all the modules that contain user data. Import from the module line items back into the same module line items
2. Then go over the Import Data Sources tab and edit the source for actions created in (1)
3. By default, the data source will point to the DEV model itself. Change this to point to the PROD model (as the same modules exist in the PROD model, editing the source will work just fine)
4. Repeat creating imports and editing import data sources for all modules which contain user inputs
5. Running the imports will then import all user data. Adding these actions into a process and then scheduling this process will ensure DEV model is always updated
Regards,
Anirudh
1 -
3: Yes it is a common practice to make copies of prod and use them as DEV, but not really for the reason you are mentioning.
Idieally you'd have your data coming into your data hub and it feeds production. it also feeds the DEV that you can update more or less frequently and with a subset of the data only.
1
