2.28 Release Notes
28 July 2025
🚀 Lakehouse Optimizer 2.28 – What's New
Version 2.28 introduces key improvements to cost visibility, incident management, and overall user experience.
🌟 Highlights
New features and functionalities
Vendor Consolidation feature has been released, with relevant data collection starting in July 2025. To learn more, see Vendor Consolidation page.
“Lakebase” cost category has been added under “Databricks Platform“ feature set on “Cost Over Time“ page.
Detailed metrics added for “Apps“ under Workloads > Cost Over Time (entity name, cost, and creator).
Detailed metrics added for “Vector Search“ and “Model Serving” under Workloads > Serverless (entity name, cost, and creator).
Detailed reporting on untagged in “Cost by Tag Group” for Creator tag.
💰 Cost & Forecasting
Improved cost accuracy with enhanced metadata and refined ingestion logic.
Stabilized cost charts and improved daily forecasting visuals.
Enhanced tooltips and cost KPIs for clearer insights.
🚨 Incidents
New Job Duration Exceeded incidents for long-running job detection.
Scoped incident rules at subscription/workspace levels for better control.
Persisted job run recommendations to support historical optimization.
🎨 UI/UX
Refined filters, tooltips, and layouts for greater consistency.
Enhanced navigation speed and dropdown responsiveness.
Improved visibility of workspace features and serverless KPIs.
⚙️ System Enhancements
Introduced serverless KPIs (hourly cost rate and data cost rate).
Enhanced DBU SKU parsing and duplicate cluster name handling.
Strengthened cluster metadata accuracy and provider tagging.
🔧 Fixes & Stability
Corrected missing metadata for clusters and SQL warehouses.
Improved API stability for search, incidents, and cost reporting.
Reduce thread pool size to 1 in consumption services for optimized resource utilization.
✅ Benefit: This release delivers stronger cost governance, enhanced performance tracking, and a streamlined experience for platform and data teams.
Forecasting // Cost & Data Usage widget
Data Usage is also known as Data Processed.
KPIs as side panel on Forecasting page
Display last 5 runs status on workflows - jobs
Navigation link "View Cost Breakdown" from Forecasting/Daily view to Forecasting/Workspaces Cost
enhanced navigation to support to investigate daily cost spikes
Vendor Consolidation
Distribution of cost per vendors
View dependent vendor query references per run
Incident to compare daily cost of job
Incident to compare daily cost of job with AVG daily cost of Last Week, Last Month.
Generate daily individual job cost incidents, when the cost of a day increases over the threshold defined in the incident rule compared to the average cost over the previous week/30 days.
For example, if today is a workday (Monday to Friday) compare to
(1) average daily cost from Monday to Friday previous calendar week (starting previous Monday) --> avg over week
(2) take previous 30 days starting with yesterday, and take average daily cost of working days in that period --> avg over month
if today is weekend (Saturday to Sunday)
compare to
(1) average daily cost from Monday to Sunday previous complete calendar week --> avg over week
(2) take previous 30 days starting with yesterday, and take average daily cost of all days in that period --> avg over month
Week starts on Monday
Real cost incidents are generated independently of estimated cost incidents
Current limitations:
when the cost of a day from the reference interval is reanalyzed/updated, the cost of the "current days" (i.e. days having the updated day in their reference interval) will not be updated
if an incident was created and the average cost of the previous days increases on further re-analyses, the incident will not be healed (deleted)
Redesigned Wokloads main menu
with individual items for
Worklflows
All-Purpose Compute
Pipelines
SQL Warehouses
Pools
Serverless
Databricks Platform
Menu has been simplified to have at the top of the pages only one group of tabs where applicable.
Tabs have the full name always fully visible.
Updated trendline kpis, reordered and included additional info
with enhanced filtering
kpi filters with multiple units selection
Export to CSV in All-Purpose Compute // Cluster instances
Support search with whitespaces on workloads
support search with whitespace in "AllPurpose/Clusters: name"
search by name with whitespace supported SQL Warehouse
support DLT pipeline search by name with whitespace
support whitespace search by name for one-time runs