Database
Vector AI
In-Memory

SAP HANA Cloud

The cloud-native evolution of SAP HANA, delivered as a managed database service on SAP BTP. Combines in-memory relational processing, vector search, document store, graph analytics, and elastic data lake in a single platform.

Overview

SAP HANA Cloud is the BTP-managed evolution of SAP HANA. Unlike the on-premise HANA 2.0 installed alongside S/4HANA, HANA Cloud is a fully managed cloud database service: SAP handles patching, upgrades, backups, and high availability. Storage and compute are decoupled, allowing elastic scaling — you can increase memory from 32 GB to 1 TB without downtime.

HANA Cloud is the primary database for CAP applications on BTP, Datasphere workspaces, and SAP Business Data Cloud. It supports multi-model data processing through specialised engines: Vector Engine (cosine similarity search for RAG and AI embedding pipelines), JSON Document Store (schema-flexible JSON documents co-located with relational data), and Graph Engine(property graphs for network analysis and relationship traversal).

HANA Data Lake extends HANA Cloud with cost-efficient cold storage. It has two variants: Data Lake Relational Engine (columnar SQL analytics on petabytes of historical data via the SAQ IQ engine) and Data Lake Files (object storage for Parquet, CSV, and ORC files accessed via HANA SQL and Spark).

HANA Cloud Database — in-memory columnar store with elastic scaling
HANA Data Lake — cost-efficient cold storage (relational + files)
Vector Engine — embedding similarity search for AI/RAG use cases
JSON Document Store — schema-flexible JSON co-located with SQL tables
Graph Engine — property graph analytics and SPARQL-like traversal
HDI containers — isolated schema-per-app for CAP BTP deployments

Multi-Model Engines

In-Memory Relational Engine

Column-store tables with row-level locking and MVCC. Supports ACID transactions at petabyte scale. Analytical aggregations run directly on compressed column vectors without materializing results.

  • Column store + row store in one engine
  • In-memory + disk-extended storage
  • ACID + MVCC transaction support
  • Native JSON, XML, and geospatial types
🧠
Vector Engine

Native cosine similarity and Euclidean distance search over REAL_VECTOR columns. Store embedding vectors from SAP AI Core / Gen AI Hub directly in HANA, enabling Retrieval-Augmented Generation (RAG) without a separate vector DB.

  • COSINE_SIMILARITY() SQL function
  • ANN (Approximate Nearest Neighbor) index
  • REAL_VECTOR column type (up to 65,535 dimensions)
  • Hybrid search: vector + SQL predicates in one query
{}
JSON Document Store

Schema-flexible JSON document collections stored and indexed within HANA Cloud. Query with SQL JSON_VALUE() / JSON_QUERY() or MongoDB-compatible API. Documents and relational tables can be JOINed in a single query.

  • COLLECTION type for JSON documents
  • MongoDB wire protocol compatibility
  • Indexing on nested JSON paths
  • JOIN with relational tables in standard SQL
Graph Engine

Property graph stored in HANA with Cypher-like GRAPH queries (openCypher via SQLScript extension). Used for supply chain network analysis, fraud detection, and hierarchical reporting.

  • openCypher graph pattern matching
  • Shortest path and reachability algorithms
  • Graph workspaces with versioning
  • Integration with SAP Analytics Cloud graph visualisation
🗄
Data Lake Relational Engine

SAQ IQ-based columnar engine for petabyte-scale cold analytics. Data ingested from HANA Cloud or external sources via INSERT INTO ... SELECT or bulk load. Cost is a fraction of HANA Cloud hot storage.

  • SAQ IQ columnar compression ratios > 10:1
  • Standard ANSI SQL with analytics extensions
  • Delta sync from HANA Cloud hot store
  • Best for historical reporting > 2 years old
📂
Data Lake Files

Object storage layer for structured and semi-structured files (Parquet, CSV, ORC, Avro). Queryable from HANA Cloud via virtual tables or directly from Spark / Databricks through HDFS-compatible API.

  • Parquet / ORC / CSV / Avro supported
  • Virtual table access from HANA SQL
  • HDFS-compatible endpoint for Spark
  • S3-compatible API for external data ingestion

HDI Containers for BTP CAP Apps

HANA Deployment Infrastructure (HDI) provides isolated, per-application database schemas on a shared HANA Cloud instance. Each CAP application deployed to BTP Cloud Foundry binds an HDI container service instance. HDI handles schema versioning, deploy/undeploy of database artefacts (tables, views, procedures), and delta migrations automatically — without custom migration scripts.

Isolation
Separate schema per app per tenant
Deployment
cf create-service hana hdi-shared
Migration
Automatic delta deploy via HDI deployer
Detailed content for this section is being added in upcoming prompt modules. Use the navigation to explore fully documented sections.