Capacity Metrics Datasets¶
Connect one or more Microsoft Fabric Capacity Metrics app semantic models to a Power BI environment, and choose which capacities MetaMiner collects from each.
How it works¶
The Fabric Capacity Metrics app is a Microsoft app that holds capacity usage (capacity units, operations, storage) in a semantic model. MetaMiner runs read-only DAX queries against that semantic model as the service principal and stores the results in the MetaMiner Database (MMDB). The Capacity usage dashboard and the capacity views in the MetaMiner Client read from there.
Each Capacity Metrics app reports the capacities its owners added to it. One environment can list several datasets, for example one app per region.
| Term | Meaning |
|---|---|
| FCM | Fabric Capacity Metrics |
| Dataset | The Capacity Metrics app semantic model, identified by its ID (a GUID) |
| Capacity | A Fabric or Power BI Premium capacity reported by a dataset |
Before you begin¶
- The Capacity Metrics app is installed and refreshed. See Capacity Metrics App.
- The service principal, or its security group, is a Contributor or Admin of the workspace that hosts the app's semantic model.
- You have the semantic model ID. Open the semantic model in the Power BI service; the ID is the GUID after
/datasets/in the browser address. - Tenant ID, Client ID and Client Secret are filled in the environment dialog, so MMA can read the capacities.
Steps¶
- Open the environment dialog: Add Power BI for a new environment, or select the environment and click Edit.
- Under FCM Datasets & Capacities, clear Ignore FCM at this time.
- Click Add, paste the semantic model ID at "Enter FCM Dataset UUID:" and click OK.
- MMA loads the capacities of the dataset. If it does not, click Load Capacities.
- Under the dataset, check the capacities to collect and clear the others.
- Repeat for each additional dataset.
- Click Apply. Validation runs one Fabric Capacity Metrics Data step per dataset; see Validate the Connection.
- Restart the server if it is running, so collection uses the new list.
Image of the FCM Datasets & Capacities tree with one dataset expanded, three capacities checked, one cleared and a DISCONNECTED group below
../../images/admin/pbi-fcm-datasets-tree.png
Input format and rules¶
| Rule | Detail |
|---|---|
| Format | One GUID per dataset, for example d5f3b2a1-1234-5678-abcd-ef1234567890. Anything else is rejected with "Invalid UUID format". |
| Duplicates | A dataset can be listed once ("This UUID is already in the list."). |
| One dataset per capacity | A capacity can be checked under one dataset only. Checking it under a dataset clears it under the others, so no capacity is collected twice. |
| First load | All capacities are checked, except those already checked under another dataset |
| No capacity checked | If you clear every capacity of a dataset, MetaMiner collects all of that dataset's capacities. To stop collecting a dataset, click Remove. |
| Remove | Select the dataset row (not a capacity) and click Remove |
What the tree shows¶
| Entry | Meaning |
|---|---|
| Dataset ID in bold | A listed Capacity Metrics semantic model |
Name (SKU, region) with a check box |
A capacity the dataset reports |
Capacity ID with N/A |
The dataset reports a capacity that the tenant capacity list does not contain, for example a removed capacity |
| Red message under a dataset | Microsoft refused the query, shown as the Microsoft error code or HTTP <code> with the description |
| DISCONNECTED group, greyed | Tenant capacities that no listed dataset reports. They cannot be selected. Add them to a Capacity Metrics app to collect them. |
Embedded SKUs EM1 to EM3 and A1 to A3 are not listed. Trial capacities (SKU starting with FT) are not listed or collected.
Include trial capacities
Set pbi.fcm.trial.capacities.enabled=true in MetaMinerServer.properties and restart the server to include trial capacities. Use this only on advice from Infolytik Support.
Turn capacity metrics off¶
Check Ignore FCM at this time and click Apply. MetaMiner stops collecting capacity metrics for the environment after the next server start. The dataset list is kept for later.
What is collected¶
For each selected capacity MetaMiner collects, per dataset in turn: the capacity list, items, metrics by item, operation and hour, capacity unit detail, and storage by workspace and hour. On the first run it reads back 14 days. Data already collected is kept for 26 weeks by default; see Data Retention.