Version 26.8.0

Date: 8th August 2026

Major Features

• Unified Tags: Tags across ADOC are now standardized as key-value pairs, consolidating the previously separate concepts of plain string tags and key-value labels into a single unified model. All tag operations — create, edit, retire, merge, and delete — are now managed from a central library under Governance > Tags. Tags can be applied to catalog assets, policies, rules, rulesets, and user-defined functions. A Data Labelers tab enables rule-based auto-classification that applies tags automatically during profiling and crawler jobs. Tag lifecycle changes (retire, merge, delete) propagate automatically across all associated entities. Existing tags, labels, and API integrations continue to work without modification — all existing data is migrated automatically on upgrade. For more information, see Tags.

• Multivariate Anomaly Detection: ADOC now supports multivariate anomaly detection in Data Anomaly Policies, enabling you to monitor related columns as a group and identify anomalies in relationships between columns that may not be detected through individual column monitoring. You can select the columns to monitor, configure the training window and sensitivity, and use either Spark or Pushdown processing. Detected anomalies appear in the standard Alerts view and can be routed through configured notification channels. For more information, see Data Anomaly Policy.

• Multi-Schedule Support for Policies: Data Quality, Reconciliation, Data Freshness, and Auto Anomaly policies now support multiple independent cron-based schedules per policy, allowing you to mirror your pipeline's actual cadence without creating duplicate policies. Up to seven schedules can be configured per policy, with a single shared timezone applying uniformly across all. Upcoming executions across all schedules are visible in the Jobs page. For more information, see [Policy Scheduling].

• Domain-Based Alert Visibility: Alert visibility is now scoped to the domains and resource groups a user is authorized to access. Users see only the alerts raised on assets within their assigned domains, eliminating cross-team alert noise in multi-team tenants. A Domain selector is available on the Alerts page for users who belong to multiple domains. For more information, see Alerts & Notifications. Important: This change affects all users in tenants where assets are organized into resource groups and domains. Users who are not assigned to a domain, or whose domain has no resource groups configured, will not see any assets, alerts, or policies — including assets they own. Administrators should verify that all users are correctly assigned to the appropriate domains and resource groups before upgrading to ensure uninterrupted access.

• Fivetran HVR Integration (Preview): ADOC now supports pipeline observability for Fivetran HVR 6, a real-time Change Data Capture replication platform. Connecting ADOC to your HVR Hub Server automatically discovers all replication channels as monitored pipelines, surfacing channel health status, capture and integrate latency metrics, throughput statistics, failure events, and source-to-target asset lineage. For more information, see Fivetran HVR.

• Visual View: Google Cloud Pub/Sub Support: Google Cloud Pub/Sub topics can now be used as source assets when building a Visual View, enabling teams to join Pub/Sub data with relational tables, files, and other streaming sources in a single unified view. The resulting Visual View can be profiled and monitored using standard Data Quality and Reconciliation policies, with Pub/Sub-specific incremental strategies (timestamp-based and lookback-based) available for ongoing execution. A connected and crawled Pub/Sub data source and DataPlane version 4.7.0 or higher are required. For more information, see 2. Visual View.

Data Quality & Policy Enhancements

• Selectable Data Quality Dimensions for Rules: ADOC now allows you to manually assign a Data Quality dimension to rules when creating or editing a Data Quality policy. This is especially useful for SQL-based rules, which default to Other but can now be classified under the appropriate dimension, such as Accuracy, Completeness, or Timeliness. You can also override the dimension assigned to other rule types when needed, helping ensure dimension-based scores and dashboards accurately reflect the intent of each rule. For more information, see Data Quality Policy.

• Uniqueness Check — NULL Handling: The Uniqueness Check rule in Data Quality policies now includes an Ignore NULL values when validating uniqueness option. When enabled, NULL values are excluded from the uniqueness calculation and multiple NULL values in a column are not counted as duplicates. The option is unchecked by default, preserving existing behavior for all configured policies.

• Rules — Advanced Filtering: The Rules tab under Governance > Policies now supports four additional filters: Tag, Asset, Column Name, and Last Execution Date. Filters support multi-select and can be combined with existing filters for precise rule lookup across large rule inventories. The same filtering enhancements are also applied to the Rules view within the Data Products Policies tab. For more information, see Rules.

• Rules — Policy ID Column: The Rules listing view now includes an optional Policy ID column, surfacing the immutable, system-assigned identifier for the policy each rule belongs to. The column is hidden by default and can be enabled through the column selector in the top-right corner of the Rules table. When combined with the rule name, Policy ID provides a globally stable, system-wide reference for any individual rule — useful for reporting, auditing, and system integrations where policy display names may change over time.

Data Products

• Data Products: Rules View: The Policies tab on the Data Product detail page now includes a Rules sub-view (Preview), providing rule-level visibility into data quality checks directly within the Data Products context. Rules are scoped to the Data Product's assets and show the rule name, asset, column(s), rule type, dimension, policy type, policy name, quality score, result, and last executed time. Users can search by rule, policy, or column name, and filter by time range, quality score, rule result, or policy type. Rules are displayed only for policies that have at least one successful execution. For more information, see Data Products.

• Interactive Asset Selection in Data Product Lineage: The Data Product Lineage page now supports asset selection, allowing you to select specific assets within a Data Product's lineage graph and view their upstream and downstream relationships in a focused, uncluttered view. This reduces the effort needed to trace a specific table or view through a dense, all-inclusive graph, making root-cause analysis and impact assessment faster.

• Configurable Asset Limit for Data Products: ADOC now allows administrators to set a maximum number of assets permitted in a single Data Product. This limit is configured at the tenant level, acting as a governance guardrail that helps keep lineage graphs performant as Data Products and data estates scale.

Data Plane, Crawling, and Storage

• Datasource-Level Crawler Resource Configuration: ADOC now enables administrators to configure crawler compute resources for individual datasource integrations. You can define memory, CPU, connection pool settings, and applicable Spark resource configurations based on the scale and requirements of each datasource. Datasource-specific settings override data plane defaults, while integrations without custom settings continue to use the existing defaults. This helps optimize infrastructure usage and improve crawl reliability without requiring data plane reconfiguration or restarts. For more information, see Configure Crawler Resources.

• Multiple Storage Buckets per Data Plane: Data Quality and Reconciliation policy results can now be directed to different S3 or GCS storage buckets within a single data plane deployment. By setting a fully qualified storage URI (s3:// or gs://) in a Persistence Configuration's Base Path field, individual policies write their results to a designated bucket, overriding the data plane's global storage setting. Relative paths continue to be appended under the global base path for sub-folder organization. All configured buckets must reside in the same region as the data plane's global storage and must be accessible with the same access credentials. For more information, see Writing to Multiple Storage Buckets.

• Improved Job Execution Reporting Reliability: Job completion reporting between the Data Plane and Control Plane is now handled asynchronously, eliminating intermittent failures where jobs appeared stuck or reported an error despite completing successfully. This improvement is most impactful in high-volume environments with many concurrent policy executions. No configuration changes or redeployment is required — the improvement applies automatically as part of the upgrade.

SDK

• acceldata-sdk-python — Unified Tag Support and Multi-Cron Scheduling: The acceldata-sdk-python package now supports Unified Tagging, allowing automation authors to create, assign, and manage key-value tags on data assets programmatically using typed, Pydantic-validated models. Policy scheduling now supports the jobSchedule.cronExpressions field, which accepts a list of cron strings, enabling a single Data Quality, Reconciliation, Data Freshness, or Auto Anomaly policy to run on multiple independent schedules with a shared timezone. For more information, see acceldata-sdk-python Overview.

UX and UI Enhancements

• Watched Assets Scoped to the Signed-In User: The Watched Assets list on the Home Page Dashboard, the Discover Assets view, and Reliability Reports now shows only the assets watched by the signed-in user, rather than assets watched by other users. Each user now sees only the assets they are actively monitoring.

• In-App Glossary Popovers for Key Product Terms: ADOC now surfaces an in-app popover when you hover over or click key product terms — such as Data Quality, Freshness, Reliability Score, and Incident — anywhere in the application. Each popover includes a plain-language explanation, a brief example, and a link to the full documentation, so you no longer need to leave the current screen to understand what a metric or concept means. This is particularly useful during onboarding or when reviewing unfamiliar terminology in reports and audits.

• Refreshed Dashboard Studio Interface: The embedded Dashboard Studio has been upgraded to its latest version, bringing a visual design that aligns with current platform interface standards. For more information, see Dashboards.

• Resolved an issue where open alerts were not consistently reflected in the Discover Assets and Manage Policies modules despite an active open incident being visible in the Alerts Listing page and Home dashboard.

• Resolved an issue where SQL-based Data Quality policy queries embedded with environment-specific asset paths were not updated during export/import migration for Databricks, causing policies to continue referencing the source environment asset paths after migration.

• Resolved an issue where the crawler scheduler incorrectly combined all configured minutes with all configured hours when multiple run times with different minute values were set, resulting in extra unintended executions beyond the scheduled times.

• Resolved an issue where reference asset mappings for policies referencing assets from a different catalog were missing from the exported package, causing policy import to fail even when the mapping entry appeared in the generated JSON file.

• Resolved an issue where a dataplane was not fully deleted when a user raised a delete request, due to a failure in handling the API key deletion response during service user credential revocation, causing the dataplane to re-register on restart with revoked certificates.

• Resolved an issue where incremental runs on Visual View policies did not propagate the incremental timestamp filter to the underlying BigQuery table, causing full table scans on every execution instead of incremental reads.

• Resolved an issue in the Azure Data Factory filter where deselecting Select All did not clear all selected data factories. Users can now easily clear their current selections and choose a different set of data factories.

• Resolved an issue where cadence jobs could not be rescheduled on a datasource after being manually disabled at the tenant level, resulting in a "Failed to acquire lock" error on the control plane.

• Resolved an issue where dashboard widgets for DQ Score by Asset Name and Policies Configured threw errors, the Top Performing Tables By Quality widget displayed duplicate asset entries, and Domain/Tag filters behaved inconsistently on dashboards.

• Resolved an issue where pushdown jobs were incorrectly marked as ERRORED when execution exceeded the workflow health check timeout, even though the underlying database query completed successfully, resulting in a misleading "Spark job completed successfully, but updating the result failed" error.

• Resolved an issue where Databricks pipeline runs could remain indefinitely in the RUNNING state because completion events were not received on Databricks Runtime 17.3 LTS. The previously documented workaround also no longer functioned. Pipeline runs now transition to their final state correctly.

• Resolved an issue where Airflow pipeline asset correlation did not support Amazon Redshift Serverless. Pipeline assets are now correctly correlated with their corresponding catalog assets.

• The add_asset_tag SDK method was not returning the tagId in its response. This occurred because the service methods (get_asset_tags and add_asset_tag) were using the AssetTagInfo model, which lacked the tagId field. The fix replaces AssetTagInfo with the AssetTag model across adoc_client.py, asset_resource.py, and asset_service.py. Response parsing was updated to use the new AssetTagResponse structure (keyed on assetTags), and both unit and integration tests were updated to validate the tagId is now correctly returned in the response.

• Resolved an issue where policies associated with a Kafka topic were archived when the topic became empty. A configuration option is now provided to prevent policy archival when a Kafka topic has no data, ensuring policies remain active and intact.

• Resolved an issue where delayed reconciliation jobs crashed with a NoSuchElementException when a policy was edited between the source-asset execution and the delayed sink-asset job, causing the rule item IDs to no longer match the stored item executions.

• Resolved an issue where the Spark engine incorrectly evaluated REGEX_MATCH DQ rules on NUMERIC/DECIMAL columns due to a missing STRING cast before pattern matching, producing artificially low and incorrect DQ scores.

• Resolved an issue where container images for certain DataPlane v3 components (ad-crawler-service, acceldata-cleanup-pods, spark-history-server-init) continued to pull from the Acceldata ECR instead of the customer-configured private image repository, causing pods to fail with 403 Forbidden errors.