Qdrant

Qdrant is a cutting-edge vector database and next-generation search engine for AI applications, available in the cloud. It offers a robust vector similarity search engine and a database solution with a user-friendly API to store, search, and manage vectors with additional payloads. Qdrant stands out for its extensive filtering capabilities, which makes it ideal for neural network-based and semantic matching, faceted search, and various other applications.

Key features

• Payload filtering and management.
  Qdrant allows you to attach JSON payloads to vectors, thereby enabling data storage and filtering by payload value. Supports various data types and query conditions, including keyword matching, full-text filtering, numerical ranges, geolocations, and much more.

• Hybrid search with sparse vectors.
  Supports hybrid search methods combining vector similarity search with traditional sparse vector search.

• Vector quantization and storage on disk.
  Optimizes storage efficiency through vector quantization and disk-based storage solutions.

• Distributed deployment and scaling.
  Offers distributed deployment options with horizontal scalability through sharding and replication, enabling zero-downtime upgrades and seamless dynamic scaling of collections.

• Query planning and load indexes.
  Uses stored load information to optimize query execution strategies to improve search performance.

• SIMD hardware acceleration.
  Uses modern processor architectures (x86-64 and Neon) for hardware acceleration delivering improved performance.

• Asynchronous input/output (Async I/O).
  Uses io_uring to maximize disk bandwidth utilization even on a network storage.

• Write-ahead logging.
  Ensures data integrity with update confirmation for protection against data losses due to power failures.

1. Create a network and a subnet.

2. Create a security group in your new network and add rules:

   Traffic direction	Description	Port range	Protocol	Source / Destination	CIDR blocks
   Inbound	Qdrant	6333	TCP	CIDR	0.0.0.0/0
   Inbound	SSH	22	TCP	CIDR	0.0.0.0/0
   Outbound	Any	All	Any	CIDR	0.0.0.0/0

3. Get an SSH key pair for connection to a virtual machine (VM).

4. Create a VM from a public image:

   1. Under Boot disk image on the Marketplace tab, select Qdrant.
   2. Under Network settings, select the network, subnet, and security group you created earlier.
   3. Under Access:
      • Enter the username in the Login field.
      • Paste the contents of the public SSH key file in the SSH key field.

   Wait for the VM to be created and save its public IP address.

5. Submit a request to create a test collection:

   curl -X PUT 'http://<VM_IP_address>:6333/collections/test_collection' \
     -H 'Content-Type: application/json' \
       --data-raw '{
         "vectors": {
           "size": 4,
           "distance": "Dot"
         }
       }'
   

   Result:

   {"result":true,"status":"ok","time":0.145993959}
   

• High-performance vector database and search.
• Hybrid search.
• Seamless integration with AI agent and process building tools such as Langflow.
• Used as a vector database for RAG.

Yandex Cloud technical support is available 24/7. The types of requests you can submit and the relevant response times depend on your pricing plan. You can switch to the paid support plan in the management console. You can learn more about the technical support terms and conditions here.