We’re excited to announce two new features to the Fabric Capacity Platform:
- Capacity platform updates with new Pause/Resume observability support for Administrators.
- Virtualized artifacts and workspaces supporting Copilot and VNET Gateway usage reporting.
Last fall we announced how the Fabric capacities platform provides the unified compute engine that powers all of the various Fabric analytics and AI workloads in a single shared SaaS experience.
One of the key pillars for the capacities team is the principle of observability for all platform and consumption experiences. Capacity Metrics delivers on this promise empowering customers to monitor and optimize usage of their capacities for efficiency, performance, and cost optimization.
Pause and resume your capacity to control costs:
Pause and Resume is a Fabric capacity platform feature that lets you manage compute costs on F SKU capacities by suspending the execution of all workloads running on the capacity. For details on how to pause a capacity please review our documentation. When a capacity administrator pauses a capacity two key things will occur on the Fabric platform:
- Workloads running on the capacity will stop execution and no longer accept new requests by users.
- The remaining cumulative overages and smoothed operations on your capacity are summed, added to your Azure bill and cleared from cumulative and smoothed usage monitored by the capacity platform. Storage costs will still accrue while capacity is in a paused state.
Observability of Pause Resume in Capacity Metrics.
In my post from last September I walked you through how the Capacities platform uses bursting and smoothing to deliver blazing performance with simplified administration. Building on these concepts, we’ve added 3 new features for pause resume support in Capacity Metrics for Fabric administrators:
- Since smoothing spreads usage up to 24 hours into the future, when a capacity is paused all future smoothed usage will be reconciled and show up as a spike in consumption in the utilization graph.
- We’ve also added a new system events table to Capacity Metrics to simplify viewing and correlation of correlation data to system pause and resume events.
- In the timepoint drill views of capacity metrics you’ll also see that the “smoothing end” fields are updated with an end time of the pause event.
Capacity Metrics for Copilot and VNET Gateways
Throughout February this year we’ve announced two exciting new analytics experiences in Fabric
- Announcing Fabric Copilot pricing | Microsoft Fabric Blog | Microsoft Fabric
- Announcing Fabric VNET Data Gateways | Microsoft Fabric Blog | Microsoft Fabric
One of the key benefits of Fabric is a common user experience unifying all the different analytics, data and AI experiences supported by Fabric. The capacities team currently drives alignment across the product division through a conformed data schema. The data contract assumes that the following fields will always be available for usage reporting, billing and chargeback allowing administrators to easily correlate capacity compute spend to an identifiable source:
- Tenant Id
- Capacity Id
- Workspace Name
- Item Name
- Capacity consumption measured in Capacity Units (CU’s)
- Wall-clock duration of operations executed
While working with Copilot and VNET Data Gateway teams we identified new requirements in their usage patterns that prevented strict compliance with these billing and usage reporting data contracts.
- Copilot code generation may be used to generate DAX for a Power BI report that isn’t physically saved to a Fabric item or may be saved by a user to a workspace item later in the user’s workflow.
- The VNET team needs to spread usage across the Capacity without mapping usage to a specific workspace or item.
Announcing new virtualized Fabric workspaces and items!
The new virtualized workspaces and items features lets customers take predictable dependencies on capacity platform data contracts to build customized solutions for:
- Governance
- Monitoring
- Chargeback
Virtualized workspaces and artifacts maintain data contracts used by our customers by allowing Fabric workload teams to populate informative virtualized values for items and workspaces that support aggregations used by both capacity metrics and our customers custom monitoring solutions. We also now keep track of whether the fields are virtualized or physically present in a Fabric capacity for full transparency by Capacity Metrics customers or programmatic consumers of capacity billing and usage data who may want to special case treatment of these experiences.
Additional resources:
Summary
We continue to evolve the Fabric Capacity Platform and observability features in Capacity Metrics to meet an ever-growing list of AI and analytics features. Pause and resume can be used to help organizations control costs and Capacity Metrics has new virtualized workspaces and artifacts that support diverse Fabric workloads while maintaining data contracts our customers build against. We can’t wait to see what you’ll build with the new features and we’re always listening to your feedback at Fabric Ideas site: https://aka.ms/FabricGovernanceIdeas