Account Registration: To get started with Upstash, you'll first need to create a free tier account. This can be done by visiting Upstash's official website and clicking on the "Sign Up" button.
Database Creation: Once you've successfully registered, your next task is to create a new vector database. This can be done by navigating to the Upstash Documentation website at https://upstash.com/docs/vector/overall/getstarted and following the steps provided there.
Dimension count
needs to be set according to the dimensions returned by the specific embedding model that is being utilized. For instance, if you're using Google's Gemini, you would be dealing with 768 dimensions. On the other hand, OpenAI's "text-embedding-3-small" operates with 1536 dimensions. This is a vital part of the setup process, as it ensures your database is correctly configured to interact with the specific embedding model you're utilizing in your operations.
Upstash will provide you with an Endpoint and Token upon the creation of the database. Be sure to make a copy of these, as you'll need to assign them to the Site Variables on your Service Center
These are essential for the end points to work.
In utilizing your newly created vector database, there are numerous key operations to become familiar with. As detailed below, these operations typically involve the addition, querying, deletion, and fetching of vectors and raw text data.
Add Vector Endpoint: This operation adds a vector to a specified index. Along with the vector, you have the option to incorporate metadata.
Add Data Endpoint: This function allows for the addition of raw text data to a specified index, after it has been embedded. Similar to the Add Vector operation, this also provides for the inclusion of metadata with the vector.
Query Vectors Endpoint: This operation makes it possible to query a given vector amidst the existing vectors within a specified index.
Query Data Endpoint: This function allows for the querying of raw text data over the existing vectors in the index after being embedded.
Delete Vector Endpoint: This operation enables the deletion of vectors corresponding to given IDs.
Clear Database Endpoint: This function is used to delete all vectors and remove the entire state associated with the index.
Fetch Vector By Id Endpoint: This operation is useful in returning data related to vectors for given, provided IDs.
By setting up a free-tier Upstash account, creating a new vector database, and learning to fully utilize the above key operations, you'll have the necessary tools to effectively implement and manage a vector-based database on the Upstash platform.