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

  1. 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.
  2. Causal Mapping: The system automatically maps relationships and causal dependencies across all telemetry signals using Smartscape.
  3. 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).

Related

More AI models by Dynatrace