Managing Planner for Query Engine Segmentation- Beta
This documentation is for Kyvos 2026.8.x. To view documentation for other Kyvos versions, click here. |
|---|
Note
This feature is applicable only to Azure Enterprise deployments with multiple segments.
1. Overview
Kyvos Query Engine Segmentation allows you to divide Query Engines into separate segments based on your requirements. You can manage each segment independently, allocating and scaling Query Engine resources to match workload requirements.
With this enhancement, Query Engine Segmentation extends beyond grouping Query Engine nodes to provide segment-level infrastructure management. The following capabilities are supported:
Scale Set-based Segmentation on Azure
Query Engine Segmentation uses Azure Virtual Machine Scale Sets (VMSS) to manage Query Engine resources for each segment. Instead of assigning individual Query Engine nodes to a segment, you can associate one or more scale sets with it. This makes it easier for the Analytical Server to manage all Query Engines in the segment.
For better workload isolation, you can create separate scale sets for different segments and map the required semantic models to those segments. Queries for each semantic model are then served by the Query Engines in the scale set assigned to its segment. This keeps the Query Engine resources for different workloads separate.
Planner-based Scheduling Management
Query Engine scheduling is managed independently for each segment. Each segment can have its own fixed schedule. Scheduling for each segment is managed through a Planner. A Planner defines the scheduling, capacity, and scaling behavior for the segments assigned to it.
Example: You can configure a planner with different capacity levels throughout the day:
For Machine type: E4s_v59:00 AM–6:00 PM: Full capacity
6:00 PM–10:00 PM: Moderate capacity
10:00 PM–9:00 AM: Zero capacity
For example, you can create separate segments for finance dashboards, ad hoc analysis, and data science workloads and configure each segment with the Query Engine and Analytical Server resources and elasticity settings appropriate for its workload.
2. What is Planner?
A Planner defines how Kyvos manages the capacity and scaling of the Query Engines associated with a segment. It determines the schedule, scaling mode, and capacity at which the Query Engines should run during a specific period.
Points to know
Existing Load based schedules must be deleted before creating planners.
You can view the Planner and Segments panes only after configuring at least one segment in Kyvos Manager on the Configuring Query Engine Segmentationpage.
If you create a planner, only the ‘Fixed Capacity’ scaling mode is supported.
The offline schedule is only supported with the default planner. The Default Planner takes precedence over all other planners. If a Default Planner and a custom planner are scheduled for the same time, the Default Planner overrides the custom planner.
If you use Query Engine Segmentation, the ‘Load-based scheduling’ option is not displayed.
Kyvos provides the following types of planners:
Planner type | Associated with | Purpose | Key capability |
|---|---|---|---|
Default Planner | Default Segment | Manage the Query Engine capacity and cluster-level scheduling | Supports Fixed/Offline/Suspend scheduling for the cluster |
Query Engine Planner | Query Engine Segment | Manage Query Engine capacity independently for a specific segment | Supports segment-level Query Engine scheduling |
2.1 When to Use What
Default Planner: Use this planner when you need to manage the entire cluster, including suspending it during periods when it is not required. For example, to suspend the cluster during weekends or weekdays for required time periods, configure an ‘Offline Schedule’ in the Default Planner.
Query Engine Planner: Use this planner when you need to manage Query Engine capacity independently for a specific segment. For example, if one workload requires full capacity during business hours while another requires a different capacity, create separate Query Engine Segments and configure their capacity independently.
2.2 Creating a Planner
While creating a planner, ensure the following:
The Default Planner associates with the Default Segment. You can assign new planners to Query Engine segments created in Kyvos Manager.
A planner can be associated with multiple segments, but a segment can be assigned to only one planner.
To create a planner, perform the following steps.
On the Cluster Scaling page, click + New Planner. The Add Planner dialog is displayed.
In Planner Name, enter a unique name for the planner.
Under Assign segments, select the required segment to associate with the planner.
Click Add to create the planner.
2.3 Configuring Planner Settings
Select the required planner.
Click the More options (⋯) icon and select Planner Settings.
Under Weekdays and Weekend, review the weekday and weekend configuration.
Under Display Time Format, select the required time format.
Review the configured Time Zone.
For Fixed Capacity, select the available capacity for Query Engines.
Click Save.
Note
Changes to the default Query Engine cluster capacity take effect from the next occurrence of the default schedule.
2.4 Assign Segments to a Planner
Assign the segments that should follow the planner's schedules and scaling configuration.
Select the required planner.
In the Segments panel, select the required segments.
Save the changes.
The assigned segments follow the schedules and scaling configuration defined for the planner.
3. Change the Planner State
Use Change State when you need to temporarily override the planner's configured schedule.
In the Active Schedule panel, click Change State.
Enter a Title for the state change.
Specify the end date and time.
Select Online under Cluster Status.
Select the required Scaling Mode.
For Fixed Capacity, select the available capacity for Query Engines.
Click Change Now.
Note
The selected state temporarily overrides the existing schedules and rules for the specified period. During this period, the existing schedules and rules are ignored.
4. Important Considerations
The following are known limitations in the current release and may be addressed in future Kyvos releases.
The audit log summary for default Planner details is currently not showing when the cluster scales up after a suspended schedule.
After creating and adding a new Query Engine Service, the segmentation details are not initialized automatically. Restart the Analytical Server to initialize the segmentation details and display them in the Kyvos Engine.
An incorrect error message is displayed when query execution fails even though an active Query Engine is available in the assigned segment.
When a schedule is executed using ‘Run Now’ with a user-created planner, the system default schedules are not executing as expected.
When the cluster is offline according to the configured Default Planner schedule, the active schedule of the applicable planner is not working as expected.
Load-based schedules created before enabling Query Engine Segmentation are not getting skipped during execution after segmentation is enabled, because load-based elasticity is not supported when multiple planners are configured.
If all segments, including the Default Segment, are already mapped to existing planners, the Default Segment continues to appear as an available option when creating a new planner.
When multiple BI Servers are out of sync, details of a deleted planner may appear in Active Schedule Details. Opening Schedule Completion Audit for a schedule associated with the deleted planner results in an error while retrieving the entity process status. Planner deletion was not synced, so deleted planner schedules may still execute as saved in the database.