Skip to main content
Question

Getting Started with Power BI Storage Questions

  • September 8, 2026
  • 1 reply
  • 8 views
Do Gooder (Customer)
Forum|alt.badge.img+3

Hey all,

We're on IFS Cloud 25R1 and working through our Analysis Models - Power BI setup (Advanced Analytics, IC10126), following Deployment Scenario 3. Getting our Azure/ADLS Gen2 side sorted with our network engineer and had a couple questions for the community:

1. Storage sizing for the Golden Workspace / ADLS Gen2
Anyone have real-world numbers or a rule of thumb for how much storage the Parquet Data Sources end up using per functional Area (Finance, HCM, Procurement, etc.)? Just trying to get a ballpark so we're not over- or under-provisioning.

Also curious if anyone's used Azure's Hot/Cool/Archive tiering on the ADLS Gen2 container to save on cost — like keeping current data Hot and tiering older Parquet snapshots down. If you've done that, did it cause any headaches with Power BI refresh?

2. Turning on just a few models to start
For our initial testing, can we enable only one or two Area models (say Maintenance and Procurement) instead of the whole set, just to keep things lean and cost-efficient while we're testing? And if we do that, is there anything we should set up differently now so it's easy to add more Areas later without redoing work?

Thanks in advance for any tips!

    1 reply

    Do Gooder (Employee)
    Forum|alt.badge.img+3

    Hi tjpatterson,

    1. Regarding sizing

    Storage sizing

    There is currently no fixed rule of thumb for storage consumption per functional area (Finance, HCM, Procurement, etc.), as the footprint depends heavily on:

    • Number of Parquet Data Sources configured
    • Volume of transactional history retained
    • Full vs. incremental load strategy
    • Number of columns included in each data source
    • Refresh frequency and snapshot retention requirements

    Parquet is a compressed columnar storage format, so storage requirements are typically significantly lower than equivalent CSV exports while also offering efficient query performance.

    • No fixed sizing recommendation per functional area.
    • No storage capacity needs to be configured upfront.
    • Customers only need an ADLS Gen2 account with Hierarchical Namespace enabled.
    • Storage consumption and cost will grow based on the number of data sources, history retained, and refresh strategy.
    • Hot/Cool lifecycle policies can be used, but any data required for Power BI refreshes should remain readily accessible.

     

    2. Turning on just a few models to start: An analysis model can be either ‘loaded on demand / manually’ or ‘scheduled’. So it’s quit flexible, and you easily start with just a few models and add more as you go.
    When importing data-sources, required by the analysis models, they will not be loaded (load data into the data lake) as long as they are not loaded via the ‘load’ button, or loaded by refreshing the analysis model. So only importing data-sources, and uploading analysis models (or publishing) will not load any parquet files into the data-lake, yet.

    Please reach out if there are more questions

    Terms & ConditionsAccessibility statement