How to add a Data Hub to existing ALM structure

Hello everyone,

 

Our current workspace/structure as follows:

DEV workspace

TEST workspace

PROD workspace

ALM: DEV --> TEST, DEV --> PROD

 

We are considering implementing a Data Hub to hold transactional data & be the single source of truth. I know it is best practice to first create the Data Hub and then the other spoke models; however, this is a 3 year old client model. I am wondering what is best practice on how to implement a data hub in this case? How would I go about ALM? How would the structure/architecture look like and what are some things to take into consideration.

 

If there is a how to article with steps and instructions, please link. Thank you!

Answers

  • @eva.cheng 

    That's quite a use case! Great question.

    Some initial thoughts - you can tell us how/what direction you want to go.

    • It sounds like the Dev-Test-Prod process is already in place WITHOUT a data hub.
    • Keep in mind ALM does NOT affect data, only structure.
    • So, it seems to me the first step is to identify the process that moves data into and out of Anaplan.
    • From there you can architect the data hub, determine what actions need to be created to replace the existing processes to import data.

    My highest recommendation for you is to start by reading about data hub performance. It's REALLY important to get off on the right process. You can very easily program yourself into a corner and make things much harder for yourself. Please read @rob_marshall best practice article on peak performance. It's a game changer, I promise.

     

    Let's keep this conversation going. It's a big use case so, take it slow.

  • @eva.cheng  Most of us go with just single data hub but a few have two data hubs one for Dev and the other for Prod.

    Anyway you will have to create Data hub first as per best practices and then push all the data thru processes to Dev. Once you validate all the processes are running fine, sample data is smoothly moving then you can think of loading full set of data into data hub and shift your focus on prod model. Now your source for prod model becomes data hub instead of anything else. So if you had any other model sending data you may want to change that to data hub by visiting to import data sources in the prod model and change it to data hub.

     

  • Hello @eva.cheng 

     

    Option 1: NO ALM to Data hub itself. Keep the data hub common for DEV/TEST/PROD. 

    Option 2: Creating a DEV/TEST/PROD data hub.

     

    In Option 1, you can create the import action in DEV from Data hub using a saved view. Once validated in DEV, sync to TEST and then to PROD.

     

    In Option 2, you have to create import action in DEV from DEV data hub using a saved view. Once validated, sync changes to Datahub TEST and Spoke TEST. Then to PROD. You need to maintain the source models, accordingly, DEV is connected with DEV data hub, TEST is connected with TEST data hub etc.

     

    Option 1 is genrally preferred.

     

    Thanks

    Arun