# Databricks cost optimization and remediation | PointFive Canonical: https://www.pointfive.co/coverage/databricks Optimize Databricks costs across clusters, jobs, storage, and ML serving. PointFive delivers DBU cost visibility, DeepWaste detection, and verified savings for Databricks environments. [All coverage](https://www.pointfive.co/coverage) # Databricks coverage Understand DBU costs, cluster usage, storage, and serverless compute in Databricks. ## Investigate Databricks usage ### Cost visibility See every dollar and what's driving it: DBU costs, compute, storage, and serverless. ### Waste detection Identify idle clusters, inefficient workloads, and unused storage. ### Remediation tracking Send findings to Slack, Jira, Teams, or ServiceNow and track each change through resolution. [Explore PointFive OS](https://www.pointfive.co/products/pointfive-os) ## Databricks coverage details | Service or resource | What PointFive analyzes | | --- | --- | | Cluster configuration | Convert all-purpose to job clusters, reduce idle time, optimize autoscale settings. | | Multi-node workloads | Identify single-threaded workloads running inefficiently on multi-node clusters. | | Storage usage | Detect unused tables and volumes, excessive Time Travel retention, redundant versioning. | | Model serving | Rightsizing recommendations for ML model serving endpoints. | | Serverless cost attribution | Allocate serverless compute costs to specific jobs, queries, and notebooks. | | Cloud and Databricks usage | Unified Databricks and cloud infrastructure savings that Databricks-only tools can't find. | - Service or resource Cluster configuration What PointFive analyzes Convert all-purpose to job clusters, reduce idle time, optimize autoscale settings. - Service or resource Multi-node workloads What PointFive analyzes Identify single-threaded workloads running inefficiently on multi-node clusters. - Service or resource Storage usage What PointFive analyzes Detect unused tables and volumes, excessive Time Travel retention, redundant versioning. - Service or resource Model serving What PointFive analyzes Rightsizing recommendations for ML model serving endpoints. - Service or resource Serverless cost attribution What PointFive analyzes Allocate serverless compute costs to specific jobs, queries, and notebooks. - Service or resource Cloud and Databricks usage What PointFive analyzes Unified Databricks and cloud infrastructure savings that Databricks-only tools can't find. [Explore the efficiency library](https://www.pointfive.co/efficiency-hub?provider-name=Databricks) ## Databricks and the cloud beneath it Review Databricks alongside your cloud infrastructure. Coordinate S3 lifecycle policies with Delta Lake retention and investigate public internet access where VPC endpoints could be used. ## Enterprise-grade security and controls Independent audits, clear privacy commitments, and controls that keep your organization in charge. Explore our certifications, security reports, and access safeguards. [Visit our Trust Center](https://trust.pointfive.co/) - [SOC 2 Type II Details](https://trust.pointfive.co/) - [ISO 27001 Details](https://trust.pointfive.co/) - [GDPR Details](https://www.pointfive.co/privacy) - [CCPA & CPRA Details](https://www.pointfive.co/ccpa) - [SAML SSO Details](https://www.pointfive.co/faq#teams-collaboration) - [Access controls Details](https://www.pointfive.co/products/pointfive-os#dataset-access) - [Penetration testing Details](https://trust.pointfive.co/) ## Frequently asked questions ### How does PointFive connect to Databricks? PointFive connects through a pre-configured notebook you run once in your Databricks workspace. It creates a read-only service principal with minimal permissions for discovery. Connect through guided onboarding. Discovery uses read-only access. Automated remediation uses customer-authorized permissions, and each customer chooses how changes are approved: per change, once for a trusted playbook, or not at all. ### What Databricks costs does PointFive optimize? PointFive optimizes cluster compute (DBUs), serverless compute, storage (Delta Lake, volumes), ML model serving, and the underlying cloud infrastructure. We find savings across every layer. ### How is PointFive different from Databricks cost tools? Databricks provides usage system tables, dashboards, and budgets. PointFive sees both Databricks and the cloud beneath it, finding cross-stack savings like S3 lifecycle coordination with Delta Lake retention that single-vendor tools can't detect. ### Does PointFive access my data or query content? For discovery, PointFive accesses metadata and usage statistics. Query text analysis is strictly opt-in and can remain disabled. Discovery uses read-only access. Automated remediation uses customer-authorized permissions, and each customer chooses how changes are approved: per change, once for a trusted playbook, or not at all. ### What security controls apply to Databricks? Discovery uses read-only access. Automated remediation uses customer-authorized permissions, and each customer chooses how changes are approved: per change, once for a trusted playbook, or not at all. Query text analysis is strictly opt-in. Dedicated service principal, minimal permissions. Metadata collection runs on isolated compute. Data encrypted in transit and at rest. SOC 2 Type II. ## Other platforms [All coverage](https://www.pointfive.co/coverage) - ### [AWS](https://www.pointfive.co/coverage/aws) - ### [Azure](https://www.pointfive.co/coverage/azure) - ### [GCP](https://www.pointfive.co/coverage/gcp) - ### [OCI](https://www.pointfive.co/coverage/oci) - ### [Kubernetes](https://www.pointfive.co/coverage/kubernetes) - ### [Snowflake](https://www.pointfive.co/coverage/snowflake) --- Source: the public page above. Product screenshots and illustrative interfaces are examples, not live customer data.