Grail
AI model by Dynatrace
The unified data lakehouse built for AI‑driven operations.
Grail is a unified causal data lakehouse for AI-driven operations that combines observability, security, and business data. It uses a massively parallel processing engine and indexless architecture to provide real-time answers from petabyte-scale data.
Model specs
The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.
Category: Data Warehouse
- Ingestion Capacity
- 1,000 TB per day
- Performance Boost
- Up to 100x vs traditional systems
- Management Overhead Reduction
- 90–99% reduction
- Data Retention
- Up to 10 years
- Query Language
- Dynatrace Query Language (DQL)
- Architecture Type
- Indexless, schema-on-read, MPP engine
- Causal data lakehouse for AI-driven operations
- Up to 100x performance boost with MPP engine
- Ingests up to 1,000 TB of data per day
- Indexless, schema-on-read architecture
- Up to 10 years of data retention
Overview
Grail is Dynatrace's unified causal data lakehouse designed to power AI-driven operations for enterprise businesses. It solves the complexity of modern cloud data by combining observability, security, and business data into a single, context-rich environment. The technology utilizes a massively parallel processing (MPP) engine and an indexless, schema-on-read architecture to provide instant access to petabyte-scale data without the need for manual indexing or rehydration. Grail serves as the foundational data layer for Dynatrace Intelligence, enabling AI agents to act with certainty based on real-time context. It supports ingestion of up to 1,000 TB of data per day and offers flexible retention periods of up to 10 years.
Data Architecture
- Unified Data Lakehouse: The causal data lakehouse unifies observability, security, and business data in a single environment for AI-driven operations.
- MPP Engine: A massively parallel processing engine searches through hundreds of terabytes of data per second to provide real-time answers.
- Indexless Architecture: The indexless, schema-on-read architecture adapts automatically to ingested data, eliminating manual schema management and re-indexing.
- Zero-latency Storage: Always-hydrated hot and cold storage ensures both live and historical data are instantly accessible without external export.
Intelligence and Analytics
- AI Foundation: Grail serves as the data foundation for Dynatrace Intelligence, fueling AI agents with reliable, context-rich data.
- Contextual Analytics (DQL): The Dynatrace Query Language (DQL) provides distributed query execution with human-readable syntax for advanced analytics.
Use cases
- SRE teams use Grail to quickly understand system behavior and respond to issues with record speed.
- Engineers unify logs and telemetry data in Grail to reduce log monitoring costs by 30-40%.
- Operations teams fuel AI agents with context-rich data from Grail for real-time intelligence and autonomous actions.
- Security teams retain security events with instant access for threat detection and incident response.
Specifications
- Access
- Available as part of the Dynatrace SaaS platform via UI and API
- Ingestion Capacity
- 1,000 TB per day
- Performance Boost
- Up to 100x vs traditional systems
- Management Overhead Reduction
- 90–99% reduction
- Data Retention
- Up to 10 years
- Query Language
- Dynatrace Query Language (DQL)
- Architecture Type
- Indexless, schema-on-read, MPP engine
- Storage Type
- Always-hydrated hot and cold storage
How it works
- Data Ingestion and Normalization: Data is ingested via OpenPipeline at scales up to 1,000 TB per day, where it is normalized and mapped to topology.
- Causal Mapping: The system automatically maps relationships and causal dependencies across all telemetry signals using Smartscape.
- Real-time Analytics: Users execute queries using Dynatrace Query Language (DQL) to get instant answers from massive datasets without pre-processing.
Pricing
- Log Analytics Ingest & Process (Pay-per-Query)
- $0.20 per GiB
- Log Analytics Retain (Pay-per-Query)
- $0.0007 per GiB-day
- Log Analytics Query (Pay-per-Query)
- $0.0035 per GiB-scanned
- Events Ingest & Process
- $0.20 per GiB
FAQ
How does Grail reduce management overhead?
Grail uses an indexless, schema-on-read architecture that adapts automatically to data, reducing manual management overhead by 90–99%.
What types of data can Grail store?
Grail unifies logs, traces, metrics, RUM data, security, and business events into a single data store with automatically mapped causal dependencies.
What is the performance of Grail's query engine?
Grail features a Massively Parallel Processing (MPP) engine that can search through hundreds of terabytes of data per second.
How long can data be retained in Grail?
Grail provides day-level retention periods for up to 10 years for stored data.
Added by AntSpace from public sources (dynatrace.com).