Alauda Data Services Vector Database E1

Introduction

Alauda Data Services Vector Database E1 deploys and manages Milvus™ vector database instances on Alauda Container Platform. It is built on the open-source Milvus Operator project, version 1.3.10, with changes for Alauda Container Platform: platform image delivery, Pod Security "restricted" defaults and an OLM package.

TOC

What it isArchitectureDeployment modesDependenciesPlatform supportVersions

What it is

Milvus is a database for storing, indexing and searching vector embeddings, for example for similarity search and retrieval-augmented generation. The operator turns a Milvus custom resource into a running database: it renders the Milvus Deployments, Services and ConfigMaps, installs the in-cluster dependencies the instance asks for, and reconciles them when you change the resource.

The operator owns these custom resources, all in API group milvus.io:

KindVersionPurpose
Milvusv1beta1One Milvus instance, in standalone or cluster mode
MilvusUpgradev1beta1A managed engine version upgrade of an existing instance
MilvusClusterv1alpha1Legacy cluster-mode API, kept for compatibility

New instances should use Milvus (milvus.io/v1beta1).

Architecture

An instance consists of the Milvus engine and three kinds of dependencies: a metadata store, object storage and a message stream (write-ahead log).

Deployment modes

  • Standalone. All Milvus functions run in one pod. Suitable for development, testing and small data sets.

  • Cluster. Milvus functions run as separate, independently scalable components. With the Milvus 2.6 engine the operator creates:

    ComponentRole
    ProxyClient entry point; serves the gRPC and REST API on port 19530
    MixCoordCombined coordinator for metadata, data, query and index scheduling
    QueryNodeLoads segments and serves searches and queries
    DataNodeFlushes, compacts and indexes data
    StreamingNodeHosts the write-ahead log streams

Clients connect to the Service <instance-name>-milvus on port 19530. Metrics are exposed on port 9091.

Dependencies

DependencyIn-cluster defaultExternal option
Metadata storeetcd, installed by the operatorAn existing etcd cluster
Object storageSilo, an S3-compatible object store installed by the operatorAny S3-compatible service
Message streamWoodpecker, the write-ahead log built into Milvus 2.6Kafka or Pulsar that you operate

Silo replaces MinIO as the in-cluster object storage for new instances. Instances that already use an in-cluster MinIO release keep it; see Upgrade.

In standalone mode the message stream defaults to Woodpecker. In cluster mode set spec.dependencies.msgStreamType: woodpecker explicitly, as the examples in Installation do; otherwise the operator defaults to an in-cluster Pulsar, whose images are not delivered with this release.

Platform support

ItemSupported
Alauda Container Platform4.1, 4.2, 4.3, 4.4
Architecturesamd64, arm64
Kubernetes1.25 or later
Milvus engine2.6.24
Install modeAll namespaces: one operator manages instances in every namespace

Versions

ComponentImageVersion
Operatormiddleware/milvus-operator:v1.3.10-acp.2Milvus Operator 1.3.10
Milvus enginemiddleware/milvus:v2.6.24-acp.2Milvus 2.6.24
Metadata storemiddleware/etcd:v1.3.10-acp.2etcd 3.5.34
Object storagemiddleware/silo:v1.3.10-acp.2Silo RELEASE.2026-09-16T00-00-00Z
Object storage clientmiddleware/mc:v1.3.10-acp.2mc RELEASE.2026-09-16T00-00-00Z
OLM packagemilvus-operator, channel stablev1.3.10-acp.2

Images are delivered through the platform image registry. You do not need to set image references in a Milvus resource: the operator fills them in from its own configuration.

No GPU engine image is delivered with this release.

Milvus™ is a trademark of The Linux Foundation. Alauda is an independent vendor. This product is not affiliated with, endorsed by, or sponsored by The Linux Foundation. All trademarks are the property of their respective owners and are used here for identification purposes only.