Perimattic Managed Qdrant - Vector Database on AWS

This AMI runs Qdrant, the open-source vector database and similarity search engine, on Ubuntu 24.04 LTS. The official Qdrant container image is pinned to Qdrant 1.20.0 (the unprivileged variant, which runs as a normal user) and baked into the AMI, so first start does not depend on a registry. Qdrant is an independent open-source project and Perimattic is not affiliated with it; product names are trademarks of their respective owners.

Why Qdrant?

Qdrant stores vectors (embeddings) with a JSON payload and searches them by similarity, with filtering, for semantic search, recommendations and retrieval-augmented generation. It has REST and gRPC APIs and a web dashboard.

What Perimattic adds

Secure by default: a Qdrant started without an API key answers every request from anyone. At first login this AMI generates a random 24-character API key and a second read-only key on your server and starts Qdrant with both. Setup then checks that anonymous and wrong-key requests are refused, that the key can create a collection, upsert points and search, and that the read-only key can read but not write, before the site is published. There are no default or shared keys in the image.

HTTPS: Apache is configured as a reverse proxy and a free Let's Encrypt certificate is requested for your domain with certbot. The REST API and dashboard are published on https://your-domain ; Qdrant itself listens only on the loopback interface.

What is and is not exposed

REST API: on HTTPS, every call except GET / (version number), /healthz, /livez, /readyz and the dashboard's static files needs the key. The dashboard page loads without a key but shows no data until you enter the key in it.

gRPC (port 6334) is not published to the internet: Apache cannot proxy it, so it listens on 127.0.0.1 only. Use an SSH tunnel for gRPC clients; the same API key applies. Only ports 22, 80 and 443 are open.

Telemetry to Qdrant is switched off.

Data and sizing

Collections are stored in /opt/app/storage and snapshots in /opt/app/snapshots on the instance's root volume (owned by the Qdrant user), so choose the root EBS volume size for your data and take EBS snapshots for backup. Qdrant held about 60 to 90 MB of memory when empty in our tests; real usage grows with your vectors (roughly 4 bytes per dimension per vector, plus index overhead, if kept in memory). A t3.medium (4 GB) is the recommended starting size; size up with your data.

Licence

Qdrant is released under the Apache License 2.0; this AMI runs the unmodified upstream image.

Getting Started Launch the AMI and open ports 22, 80 and 443 in the security group SSH in and answer the setup prompts (domain and admin email) Copy the API key printed at the end of setup Call https://your-domain/collections  with the header api-key:

A domain name with an A record pointing at the instance is required, because the API is served over HTTPS. Embeddings are created by your application; Qdrant does not generate them.

This AMI is built and supported by Perimattic.

Book a free setup consultation at https://cal.com/gaurav-pareek-perimattic/marketplace-setup-consultations  or email us at aws-support@perimattic.com .

Support Managed and supported by Perimattic, a cloud infrastructure company with 13+ years of experience serving global clients. We offer free setup assistance, custom integrations, infrastructure consulting, and 24/7 support.