3.0 Release Notes

3.0 Release Notes

Dec 31, 2025

 


Lakehouse Optimizer 3.0 is a major release that delivers a substantially expanded Optimization Review experience, new incident and recommendation types, a redesigned Usage Distribution visualization with percentile-based metrics, and broad improvements to Forecasting, Recommendations, and Workspace Cost pages. This update also introduces Unity Catalog agent storage for DBFS-disabled environments, enhanced consumption billing support for DBCU plans and Reserved Instances, and significant performance optimizations across the platform.


image-20260218-150732.png
image-20260218-150754.png

 

Optimization Insights

  • Launched Optimization Review (beta) with comprehensive before/after comparison for Jobs, Pipelines, SQL Warehouses, and Notebooks, including new APIs for each workload type

  • Added Data Usage statistics (per-run and daily/annualized) to the Jobs summary and details panels, with a new Data Unit Cost KPI panel

  • Introduced median vs. average toggle for cost and duration statistics across all Optimization Review views

  • Added pattern-matching job search across all workspaces and refactored filters for improved UX with persistent selections

  • Redesigned Daily/Annualized view with tabs, moved "What's Included" to the top, and added clickable navigation links from summary to detail views

  • Fixed permission issue preventing non-admin users from searching by job name, and resolved multiple workspace service request duplication

 

Workflows & Operations

  • Added summary KPIs for all run types and scaling of data units (KB/MB/GB/TB) on job runs pages

  • Enabled prefix search for job run IDs and job IDs in workflow runs filter

  • Added cluster instance ID linkage for the autoscaling timeline

  • Improved performance for workflow data pipeline processing and pipeline workflow run cost analysis

 

Spend Insights & Forecasting

  • Saved number of tasks in workflow run cost analysis table and refactored daily data availability logic with migration to the new table

  • Added cutoff date support for Forecasting Cost & Data Usage charts and display of real Databricks cost vs. VM estimated cost in job runs

  • Added forecaster to the same Docker network as LHO for deployment integration

  • Fixed Forecast month exceeding chart boundaries and monthly commit line overflow

  • Fixed Forecasting tags-report endpoint performance taking 20+ seconds per tag

 

Optimization Guidance

  • Added configurable date picker on the second page, category in URL for deep linking, and banner on the main screen

  • Added "created at" and "last update" timestamp columns for incident policies in the Recommendations view

  • Updated "View Cluster Instance Activity Histograms" link to "View Cluster Instance Usage Distribution" and repositioned "Open in Trends"

  • Added application ID display as a column in cluster instance from runs

  • Fixed suboptimal spend showing old values when changing period and columns filter minimum selection count

  • Updated Recommendations info text on Runs page, fixed N/A impacted Tag Groups display, and corrected "Open in Databricks" option availability

 

Governance & Alerts

  • Added SQL Warehouse Auto Shutdown Timeout incidents and interface support for Auto Stop missing and above threshold

  • Introduced Databricks Platform SKU cost incidents for monitoring platform-level cost anomalies

  • Added new incident and recommendation type for jobs with high cluster setup overhead

  • Added support for custom percentiles when generating over/under/imbalanced-provisioning incidents with updated root cause formatting

  • Added tracking for incident policy creation, updates, and deletions

  • Updated Incident Policies page styling with new tab design, improved menu color highlighting, and fixed incident email opt-out link

  • Fixed user role permissions preventing non-admins from configuring or deleting incident policies

  • Fixed SQL Warehouse incident action not opening the workload, wrong date selection navigating from incidents, unit dropdown showing when only one unit is supported, and filter reset on close

 

Resource Usage Visibility

  • Introduced percentile-based CPU/Memory usage distribution visualization with single executor, cluster-wide, and merged views

  • Added configurable max/median/average displayed values for the Cluster Instance page and CPU/Memory utilization across all executors

  • Added link to cluster instance from autoscaling timeline

  • Migrated APC clusters tab to UTC date handling

  • Added automatic cluster ownership update when the original creator no longer exists in the workspace

  • Fixed Activity Histograms tooltip missing in merged chart for over 20 executors and displaying values from the previous point

  • Fixed Usage Distribution single executor view defaulting to wrong executor, cluster view tooltip mismatch, and start/end time display to include the day

  • Fixed cluster instance response for runs removed from Databricks and excessive shutdown timeout recommendations

 

image-20260218-151116.png

 

Cost Visibility

  • Added cross-page navigation between Cost Over Time and Workspace Cost by selected resource and provider tab

  • Added sorting direction indicators on Workspace Cost resource table with default DESC sorting from Cost Over Time navigation

  • Fixed missing graphic when navigating to Cost Over Time from Workspace Cost and CSV export including wrong workspace data

  • Fixed CSV export time interval mismatch with selected date range and VM Disks naming in export

 

image-20260218-152102.png

Serverless

  • Implemented serverless Jobs trends view with cost and run analytics

  • Enhanced serverless jobs query history data loading performance

 

Billing & Consumption

  • Added Azure Reserved Instance (RI) plan discount and cost resolution support

  • Added Databricks unit plan usage tracking during Azure consumption loading and inferred discount handling for legacy pricing

  • Added new cost categories: AI Functions, Agent Bricks, Base Environments, Data Quality Monitoring, and Delta Sharing

  • Added ability to retrieve all Databricks charges from billing usage and display Modern vs. Legacy on the consumption page

 

Deployment & Setup

  • Added support for storing LHO Agent in Unity Catalog when DBFS is disabled, with removal of Workspace Files init script support

  • Added support to store and display DBFS status of Databricks workspaces

  • Added volume init script overrides for environments requiring permission grants

  • Pushed deployment scripts archive to blob storage on release

  • Fixed global init script not being disabled on init script failure

  • Fixed search in Workload Provisioning for All-Purpose Compute and page size reset issues

  • Fixed agent queue client being called synchronously during startup, causing cluster start failures

  • Updated Workspace Provisioning default page size to 100 and improved tooltip descriptions for deployment alerts

  • Fixed provisioning loading state showing previous workspace while loading and non-admin access to provisioning pages

 

image-20260218-152159.png

Navigation & Usability

  • Added timezone (UTC) badge to all date pickers across the application

  • Added NProgress loading bar for page transitions

  • Implemented daily LHO User Activity Report scheduling

  • Removed deprecated "Cluster Instances Activity Histogram" menu link

  • Fixed Settings navigation menu highlighting wrong item and datepicker highlighting wrong option

  • Fixed empty tooltip on Trend view and embedded discount plan information in tooltips

 

Reporting & Visual Clarity

  • Updated chart X axis to show units for 1st, 10th, and 20th of each month and rotated Y axis labels for consistent direction

  • Abbreviated numbers over 1000 on Pools main pages (e.g., $1,345 displayed as $1.35k)

  • Added duration and count values displayed with comma separators for day units

  • Fixed chart data points not ordered by time and display of data usage in dynamic sizing

  • Fixed cluster instance graph "load more" behavior and cluster trends filter button state after refresh