Docs
Requirements

Deployment Requirements

In this section, we will cover the requirements for deploying the project.

Software Requirements

  • An TiDB Serverless Vector (opens in a new tab) account.
    • Currently only TiDB Serverless tier has the support for the Vector Search. You can use the free tier with 25GiB storage. We suggest to choose a nearby region to reduce the latency.
    • TiDB will probably introduce the Vector Search feature to the open source version next quarter, so stay tuned.
  • Docker Compose, you can install it from here (opens in a new tab).

GenAI Services / API Keys

  • An OpenAI API key, you can get it from here (opens in a new tab). It will be used for these purposes:

    • LLM(Large Language Model) for knowledge graph extraction and chat engine.
    • Embedding model for converting text into vectors.
    • Chat Engine for generating the answer for the question asked by the user.
  • A Jina AI API key, you can get it from here (opens in a new tab), it is free for 1M tokens. It will be used for reranking the results retrieved from vector storage.

  • An LangFuse API key, you can get it from here (opens in a new tab), it is used for tracing the RAG application to debug and optimize the performance.

Web Hosting

Hardware

You can use any of the following web hosting services to deploy the project:

We suggest the following configuration for the server:

NameValue
Memory8 GB RAM
CPU4 vCPUs
Disk100 GB SSD
Number of servers1

Currently the corpus type is text, so 100 GB SSD is enough for most of the cases, but if your corpus is large then you may need more storage.

Domain name - Optional

You need to have a domain name to deploy the project on the server. You can get a domain name from GoDaddy (opens in a new tab), Namecheap (opens in a new tab), Google Domains (opens in a new tab), etc.

SSL certificate - Optional

You need to have an SSL certificate to secure the communication between the client and the server. You can get a free SSL certificate from Let's Encrypt (opens in a new tab).