Qdrant on Azure Kubernetes Service

Add vector similarity search to your applications without sending your data to a hosted service. You get Qdrant 1.19.2, an open-source vector database (it stores embeddings, the numeric fingerprints AI models produce for text or images, and finds the closest matches), installed from the Azure portal as an Azure Kubernetes Service (AKS) cluster extension. Your vectors and their metadata stay in your own subscription and network.

Who it's for

• AI and application teams building retrieval-augmented generation (RAG) or semantic search on Azure
• Platform teams that want a self-hosted vector database next to their workloads on AKS
• Organizations that must keep embeddings and metadata inside their own tenant

What you can use it for

• Retrieval for RAG chat assistants: store document embeddings and fetch the closest passages for each question
• Semantic search over documents, support tickets, or product catalogs
• Similar-item lookups for recommendations or duplicate detection

What's included

• Qdrant 1.19.2 on a minimal container image built on Wolfi (a security-focused Linux base); the release binary is checked against a checksum pinned in our build
• REST API (port 6333) and gRPC API (port 6334), both internal to the cluster; gRPC is a binary protocol used by Qdrant client libraries
• Payload filtering: combine similarity search with filters on the metadata (payload) stored with each vector
• Persistent storage: an 8 GiB volume (a PVC, Kubernetes' claim on storage that outlives the pod) through your cluster's default storage class (an Azure managed disk on a standard AKS cluster) for your collections and the write-ahead log (the journal Qdrant replays after a restart)
• Anonymous usage telemetry turned off

After you deploy

1. Choose a new or existing AKS cluster in the Azure portal wizard and create the extension.
2. Check the server: run kubectl port-forward to port 6333 of the Qdrant service and open /readyz.
3. Connect your applications from inside the cluster to the REST (6333) or gRPC (6334) endpoint and create your first collection. No API key is needed: this release turns on no authentication.

How it fits Azure

• Installs as an AKS cluster extension, so it is listed and managed alongside your cluster's other extensions
• Pay-as-you-go software charge, billed on your Azure invoice alongside your AKS costs

Security

• Runs as a non-root user (UID 1000)
• Container image pinned by SHA-256 digest (a fingerprint that guarantees you run the exact build we published)
• No public endpoint is created; the service is internal to the cluster (ClusterIP) until you choose to expose it
• No authentication or TLS (encryption in transit) is turned on: any workload that can reach the service can read, write, and delete collections. Put an authenticating gateway with TLS in front before anything outside the cluster connects

Good to know

• Runs a single Qdrant instance, not a distributed cluster, so it is not a high-availability deployment
• Default pod limits are 1 vCPU and 1 GiB of memory, with an 8 GiB volume
• Snapshots (point-in-time exports of a collection) are written to temporary pod storage and are lost when the pod restarts; download each snapshot you want to keep
• No web dashboard is included; use the API or a Qdrant client library
• Requirements: an AKS cluster with a default storage class; new clusters from the wizard use Standard_B4ms, Standard_DS2_v2, or Standard_D4s_v3 nodes

Licensing and trademarks: Qdrant is open-source software under the Apache-2.0 license. Packaged and supported by Derek Coleman & Associates Inc. This offer is not affiliated with or endorsed by Qdrant Solutions GmbH or the Qdrant project.

Support: quick start and contact details at https://www.dcassociatesgroup.com/support/qdrant-aks/

Offer type: KubernetesApps.