[Michael Flanakin (flanakin)]

🛠️ Description

🤖 [AI] Lets users turn on Power BI incremental refresh for the storage reports' Costs table. Today the incremental refresh dialog can't detect a usable column, because x_IncrementalRefreshDate is stored as text and filtered inside a helper function.

• Design: ftk_Storage now types x_IncrementalRefreshDate as Date/Time (midnight of the charge date) instead of text, and the storage Costs query ends with a Table.SelectRows on RangeStart/RangeEnd (>= start, < end) so the incremental refresh dialog can detect the filter (Power BI storage reports: incremental refresh isn't detectable out of the box #2314, discussion Incremental refresh CostSummary.storage #1799).
• ⚠️ Breaking change: x_IncrementalRefreshDate changes from text to Date/Time in all storage tables (Costs, Prices, ReservationDetails/Transactions/Recommendations). Custom filters, measures, or visuals that treat it as text or a number need updating. No new column.
• Filter behavior: When RangeStart/RangeEnd are null the filter passes every row; when set it matches the ftk_Storage row filter (whole days, same null handling).
• Scope: Incremental refresh is set up for Costs only. Other tables only get the type change.
• Untested in Power BI Desktop. Tester should check:
  1. Params empty: Costs row count and totals match dev.
  2. Set RangeStart/RangeEnd to a month: Costs only has that month; x_IncrementalRefreshDate shows Date/Time.
  3. Right-click Costs > Incremental refresh: slider is enabled; apply a policy, publish, refresh in the service; no duplicate or missing days across partitions.

Milestone: v16 (breaking change, and the Desktop/service test didn't happen before v15).

Fixes #2314

📋 Checklist

🔬 How did you test this change?

• 🤏 Lint tests
• 🤞 PS -WhatIf / az validate
• 👍 Manually deployed + verified
• 💪 Unit tests
• 🙌 Integration tests

📦 Deploy to test?

• Hubs + ADX (managed)
• Hubs + Fabric (manual) — URI:
• Hubs (manual)
• Hubs (no data)
• Workbooks
• Alerts

🙋‍♀️ Do any of the following that apply?

• 🚨 This is a breaking change.
• 🤏 The change is less than 20 lines of code.

📑 Did you update docs/changelog.md?

• ✅ Updated changelog (required for dev PRs)
• ➡️ Will add log in a future PR (feature branch PRs only)
• ❎ Log not needed (small/internal change)

📖 Did you update documentation?

• ✅ Public docs in docs (required for dev)
• ✅ Public docs in docs-mslearn (required for dev)
• ✅ Internal dev docs in docs-wiki (required for dev)
• ✅ Internal dev docs in src (required for dev)
• ➡️ Will add docs in a future PR (feature branch PRs only)
• ❎ Docs not needed (small/internal change)

🤖 Generated with Claude Code

[Roland Krummenacher (RolandKrummenacher)]

The logic looks right. x_IncrementalRefreshDateTime converts the OLE serial correctly, and the filter mirrors ftk_Storage, so nothing changes when RangeStart/RangeEnd are empty.

1. As the description says, this hasn't been tested in Desktop or the service, and no CI job builds the TMDL. Can someone run the three test steps before merge? If not for v15, v16 sounds right.
2. Rows with a null x_IncrementalRefreshDateTime pass the filter in every partition. With an incremental refresh policy, such rows would be loaded once per partition and inflate costs. Consider dropping them when the range is set, or document it.
3. Nit (setup.md): "If you need to configure incremental refresh, consider using FinOps hubs…" now sits right after the new incremental refresh link. Maybe "For larger datasets, consider FinOps hubs…".

[Roland Krummenacher (RolandKrummenacher)]

Points 2 and 3 from my earlier review are addressed in 502853a (null-date rows are documented, and the setup.md wording is clarified). Point 1 is still open: nobody has run the Power BI Desktop and service test yet, and I haven't either. Following the rule in the description ("If testing can't happen in time, move this to v16"), I suggest moving this to v16 unless someone can run the three test steps before the v15 release.

[Michael Flanakin (flanakin)]

🤖 [AI] Roland Krummenacher (@RolandKrummenacher) Thanks. Point 1: moved to v16, since the Desktop/service test didn't happen before v15. While at it, dropped the new x_IncrementalRefreshDateTime column and changed the existing x_IncrementalRefreshDate from text to Date/Time in place (ftk_Storage and all storage tables). The final Table.SelectRows on RangeStart/RangeEnd stays so the incremental refresh dialog detects it. Changelog moved to Unreleased and marked as a breaking change; setup.md and the data dictionary updated. Points 2 and 3 were already addressed. The three Desktop/service test steps still need to run before merge.

[Roland Krummenacher (RolandKrummenacher)]

Re-reviewed 0a2d198. Changing x_IncrementalRefreshDate to Date/Time is cleaner than a second column. The type is consistent across Costs, Prices, ReservationDetails, ReservationRecommendations, and ReservationTransactions. The ftk_Storage filter and the Costs filter use the same comparison, and the changelog entry moved to Unreleased with the breaking change marker, which resolves my changelog thread.

Approving. Please still run the three test steps in Power BI Desktop and the service before v16 ships.

[Roland Krummenacher (RolandKrummenacher)]

Re-reviewed 2e2f059 after the rebase onto dev. The net diff matches what I approved at 0a2d198 except for ms.date. Still approved.

[Roland Krummenacher (RolandKrummenacher)]

Re-approving after the rebase onto dev. The range-diff against my last reviewed head (2e2f059) shows the commits are unchanged apart from ms.date bot commits. The red Pester check is the broken finops-toolkit-ecosystem.md link in roadmap.md on dev, which #2341 fixes. This PR didn't cause it.