Acceldata Data Observability

Acceldata data testing

Acceldata runs rule-based checks on SQL tables, reconciles tables across sources, and shows the rows that fail.

• SQL Table Tests: Data Quality policies include null checks, uniqueness, volume thresholds, schema drift, and custom SQL/Python/JavaScript rule expressions run against SQL tables (docs).

• Table Comparison Tests: Reconciliation policy type compares data across systems with row count validation and smart cross-system matching to detect mismatches between source and destination (website).

• Inspect Failed Rows: Policy execution labels records as 'Good and Bad records'; REST API supports 'Data Preview & Validation — retrieve sample rows to verify data correctness' (docs).

• DataFrame Tests: The Python SDK is a control-plane SDK for managing policies and pipelines; no Pandas, Spark, or Polars DataFrame validation library is provided (docs).

• Unstructured Data Tests: No LLM-based validation of free-text column content found; AI features focus on anomaly detection and root cause analysis, not semantic validation of unstructured data (docs).

• Data Diff: Reconciliation uses 'smart matching' across systems; unclear whether it surfaces individual inserted/deleted/changed records vs. aggregate-level mismatch detection (website).

• Data Contracts: No evidence of machine-readable YAML/JSON data contracts formalizing producer-consumer obligations found in docs or product pages (docs).

Acceldata data observability

Acceldata profiles data and derives dynamic bounds from history to flag anomalies and drift. Monitors are configured in the UI, and Acceldata also monitors pipelines across Airflow, dbt, Spark, AWS Glue, Azure Data Factory and Fivetran.

Compare: data observability tools · data profiling tools

• Data Profiling: Asset profiling computes null %, distinct count, min/avg/max, standard deviation, and value distributions per column; supports full, selective, and incremental modes (docs).

• Anomaly Detection: Profile Anomaly and Data Drift policies derive dynamic upper/lower bounds from historical profiling data and flag statistical deviations; sensitivity level controls (low/medium/high) supported (website).

• Freshness Monitors: Data Freshness policy type monitors data update frequency (event-triggered hourly) and checks whether tables are updated within their expected ingestion window (docs).

• Volume Monitors: Dedicated rules for 'Change in Asset Size' and 'Absolute Row Count' detect unexpected changes in dataset volume (docs).

• Schema Monitors: Schema Drift policy type detects column additions, removals, renames, and type changes; event-triggered after each crawl (docs).

• Field Monitors: Asset profiling records column-level metrics (mean, stddev, null rate, distinct count) as time series across profiling runs; Profile Anomaly policies flag deviations from historical baselines (docs).

• Distribution Monitors: Profile Anomaly and Data Drift policies track column value distribution changes over time using statistical bounds computed from historical profile runs (docs).

• UI-Based Monitors: No-code and low-code policy editor in the ADOC UI; policies are configured through the Asset Details page with a graphical interface requiring no YAML or code (docs).

• Pipeline Monitoring: Monitors workflows across Airflow, dbt Core/Cloud, Apache Spark, AWS Glue, Azure Data Factory, Fivetran, and AutoSys via OpenLineage and SDK integrations; tracks job execution time, status, and span-level timing (docs).

• Monitors-as-Code: Policies can be exported/imported as a ZIP configuration package across environments, but no YAML/Python/JSON file format for version control or CI/CD deployment was confirmed (docs).

• Segment Monitors: No evidence of group-by or per-dimension monitor breakdowns found in documentation or product pages (docs).

• Custom SQL Monitors: Custom SQL, Python, and JavaScript rules exist in Data Quality policies and can be scheduled, but no evidence that returned values are tracked as time-series metrics with anomaly detection (website).

• Partition-Based Monitors: No evidence of partition filtering or time-filter scoping to a recent data slice found in documentation (docs).

• Window-Based Monitors: No evidence of tumbling or sliding time-window metrics or intra-day bucketing found in docs or product pages (docs).

Acceldata incident management

Acceldata has a full incident workflow: alerts, triage, feedback, resolution, ticketing and on-call integrations.

• Alert Notifications: Real-time alerts via email, Slack, Microsoft Teams, and webhooks when checks or monitors fail (docs).

• Incident Triage: Alert lifecycle includes Open, Acknowledged, In Progress, Dismissed, and Resolved states with mandatory audit comments; supports bulk status changes and history tracking (docs).

• Alert Feedback: Users can mark alerts as relevant or irrelevant (false positive) and snooze recurring alerts for 24 hours; feedback is used to improve anomaly detection recommendations over time (docs).

• Incident Resolution: Alerts can be marked Resolved with comments; History tab logs all past occurrences and status changes for root-cause tracking and audit trail; system auto-resolves when monitor re-runs successfully (docs).

• Ticketing Integrations: Native integrations with Jira (API token and OAuth) and ServiceNow (standard and OAuth) for creating and syncing incident tickets from ADOC alerts (docs).

• On-Call Integrations: PagerDuty and Opsgenie are listed in the Incident & Alert Management integration category on the integrations page (website).

Acceldata data quality analytics

Acceldata's dashboards and health scores track data quality.

• Data Quality Dashboards: Pre-built Policy Summary dashboard shows platform-wide policy execution and quality coverage; Alerts dashboard shows severity trends; custom dashboards available via widget library (docs).

• Data Health Scores: Each asset displays a 'policy compliance score'; Reliability Reports rank assets by their lowest average scores and show reliability score trends over time (docs).

• Incident Resolution Metrics: Acceldata markets outcomes like "minimize MTTR by up to 90%" in case studies/blogs, but no accessible documentation confirms a dedicated in-product dashboard tracking MTTD/MTTR trends or resolution rate by team/owner; the Reports doc page returned a 404 (website).

Acceldata data lineage

Acceldata infers lineage at table and column level from connected systems, with OpenLineage support and impact analysis.

Compare: data lineage tools

• Table-Level Lineage: Table-to-table lineage tracked from query logs, OpenLineage integration, and transformation manifests across Snowflake, BigQuery, Databricks, and other sources (docs).

• Column-Level Lineage: Column-level lineage traces field-to-field dependencies with data type and quality score visibility down to individual columns (docs).

• End-to-End Lineage: Lineage spans from Fivetran/AWS Glue ingestion through dbt/Spark transformation to Tableau and Power BI consumption layers (docs).

• Impact Analysis: Lineage tab identifies which downstream assets (reports, dashboards, datasets) are affected by upstream failures or schema changes (docs).

Acceldata data governance

Acceldata's catalog shows reliability scores, alerts and policies per asset.

Compare: data catalog tools · data classification tools

• Data Catalog: Discover Assets lists assets from all connected sources with reliability scores, alerts, tags, and applied policies, and a Business Glossary links glossary terms to assets and data products (docs).

• Data Catalog Integration: Pushes data reliability scores and quality metadata into Atlan and Alation; AWS Glue is also listed as a metadata integration (docs).

• Data Classification: Data Labelers apply tags automatically when a regex matches an asset's name or its column values, running during crawl and profiling jobs; the pricing page lists Automated Data Classification in Pro (docs).

• Data Products: Acceldata's "Domains" feature groups assets/reports for access control but documentation explicitly excludes monitors/policies from that grouping; a separate "xDP" data-products platform exists but its docs describe data-plane/storage architecture without confirming monitor/quality-score grouping by data product (docs).

Acceldata FinOps

Acceldata tracks query performance and cloud compute spend, with chargeback and showback by team.

• Query Performance: Query Studio provides SQL query execution metrics and query optimization identification for Snowflake and Databricks workloads (pricing).

• Cost Monitoring: Cost Metrics & Trends track cloud compute spend with Chargeback and Showback attribution by team or pipeline; cost spike alerts are included (pricing).

Acceldata LLMOps

Acceldata traces AI agent execution chains and LLM calls, with prompt versioning and evaluation management.

• AI Observability: Monitors AI agent execution chains and LLM tracing with prompt versioning, evaluation management, AI guardrails, and production failure analysis (website).

Acceldata workflows

Acceldata has a REST API, a Python SDK, webhooks and a dbt integration. It has no MCP server.

• Scheduling: Data Quality policies support scheduled automatic execution; Data Freshness policies run on hourly event triggers; scheduling is a first-class configuration option per policy (docs).

• Webhooks: Real-time HTTP callbacks on check/monitor failure via webhooks, listed alongside Slack, email, and Teams as a notification channel (docs).

• REST / GraphQL API: Comprehensive REST API for asset discovery, data preview/validation, reliability monitoring, governance enforcement, and access control management (docs).

• Python SDK: Python SDK provides programmatic access to pipeline registration, asset management, and policy creation/management APIs; an Airflow SDK is also available (docs).

• dbt Integration: Integrates with dbt Core via OpenLineage and dbt Cloud via datasource integration for pipeline monitoring and lineage tracking (docs).

• MCP Server: No Model Context Protocol server documentation found; no mention of Claude, Cursor, or Goose integration in docs or integrations page (docs).

• CI/CD Integrations: No GitHub Actions, GitLab CI, or pipeline-blocking integration documentation found; REST API could be used for custom CI/CD but no native integration was confirmed (docs).

Acceldata AI agents

Acceldata's agents work through a chat interface: they write checks from prompts, suggest checks, triage issues and apply fixes within approved policies, with a human in the loop. It also traces LLMs and AI agents.

• AI Chat Assistant: ADM provides an in-product conversational interface for natural language data management — users chat with AI to configure policies, query data health, and execute workflows (docs).

• Natural Language Test Generation: Acceldata's Policy Intelligence capability has a conversational interface letting users "ask about policies or create them in plain language," generating the corresponding data quality policy/rule from the prompt (website).

• Automated Test Generation: AI-based automatic rules are derived from asset type detection without user prompts; AI Copilot generates quality policies from data profiles automatically (website).

• AI Issue Triaging: Autonomous agent walks lineage, correlates signals from multiple sources, and isolates root causes — AI reasoning about WHY failures occurred, not just lineage graph traversal (website).

• AI Issue Resolution: AI agents propose and execute fixes within approved policies via HITL workflows; 'the fix is applied — or proposed for human approval' with every action logged (website).