AlloyDB Omni codelabs

Explore hands-on codelabs for AlloyDB Omni and AlloyDB AI to learn how to deploy and manage databases, run local AI models, and perform vector search. You can also learn how to build retrieval-augmented generation (RAG) applications and integrate AI agents by using Model Context Protocol (MCP).

Note: Some of the codelabs on this page provision managed AlloyDB for PostgreSQL clusters on Google Cloud to demonstrate AlloyDB AI capabilities and MCP integrations.

Deployment and operations

• Deploy AlloyDB Omni on a single Compute Engine VM

  Provision a Compute Engine virtual machine (VM) and install AlloyDB Omni by using RPM packages.

• Deploy and manage AlloyDB Omni on Kubernetes

  Install the AlloyDB Omni Kubernetes operator on a Google Kubernetes Engine (GKE) Autopilot cluster or your own Kubernetes cluster and manage database lifecycle operations.

• Deploy an AlloyDB Omni high availability cluster on Compute Engine VMs using the RPM orchestrator

  Deploy a high availability (HA) AlloyDB Omni cluster on Compute Engine VMs by using the RPM orchestrator.

• Logical replication with AlloyDB Omni

  Deploy multiple AlloyDB Omni instances on a Compute Engine VM and configure logical replication between them by using the pglogical extension.

• Back up and restore an AlloyDB Omni RPM deployment to S3-compatible storage by using pgBackRest

  Configure pgBackRest to back up and restore an AlloyDB Omni RPM deployment by using S3-compatible object storage.

• Back up and restore AlloyDB Omni to S3-compatible storage by using the RPM orchestrator

  Back up and restore an AlloyDB Omni high availability (HA) cluster to S3-compatible object storage by using the RPM orchestrator.

• Enable transparent data encryption (TDE) on an AlloyDB Omni RPM deployment

  Configure and enable transparent data encryption (TDE) on an AlloyDB Omni RPM deployment by using File KMS, and verify disk-level data encryption.

• Accelerate analytical queries with the columnar engine in AlloyDB Omni

  Deploy AlloyDB Omni on a Compute Engine VM, load data, and use the columnar engine to accelerate analytical queries.

Local and open AI models

• AlloyDB Omni and EmbeddingGemma with Gemma 4 on Kubernetes

  Deploy AlloyDB Omni, EmbeddingGemma, and Gemma 4 on GKE, register the models in AlloyDB Omni, and run vector search and large language model (LLM) queries from SQL.

• Advanced AI with Gemma 4 and AlloyDB Omni for regulated industries

  Run AlloyDB Omni with open-weights models such as Gemma 4, TimesFM, and TabFM in disconnected or regulated environments.

Vector search, AI functions, and RAG

• Get started with vector embeddings in AlloyDB AI

  Generate vector embeddings, create vector indexes, and run similarity searches by using AlloyDB AI and Vertex AI.

• Generate vector embeddings at scale with AlloyDB for PostgreSQL

  Backfill vector embeddings in batches, set up incremental triggers to automatically embed new rows, and run hybrid vector and SQL searches by using AlloyDB AI.

• AlloyDB for PostgreSQL RAG application with MCP Toolbox for Databases

  Configure MCP Toolbox for Databases with AlloyDB and build an interactive RAG application.

• AlloyDB AI operators and reranking

  Use AlloyDB AI operators for semantic search, semantic joins, and result reranking in SQL queries.

Agents and MCP

• Use AlloyDB Omni with MCP Toolbox for Databases for agent integration

  Connect AI agents to AlloyDB Omni by using MCP Toolbox for Databases to inspect schemas, execute SQL, and analyze query performance.