Skip to main content

Key-value store

Store arbitrary key-value data in a table.

Tableland is all about relational data between tables. But, relational databases also make excellent key-value stores. When you store key-value pairs in a relational database, your keys become relational links to other content and data, adding nice query features that SQL was designed for.

Setup​

Let's say you want to store key-value pairs that map strings to JSON data. You'll need to create a table with two columns, specifying that your key column (e.g., k) is of type TEXT and PRIMARY KEY, and your value column (e.g., v) is of type TEXT.

CREATE TABLE key (
  k TEXT,
  v TEXT,
  PRIMARY KEY (k)
);

Tableland also supports the BLOB (binary large objects) datatype. Certain use cases might prefer to use this as a type for the value v since the data is stored exactly as it was input (e.g., as bytes).

note

The word key is a reserved keyword in SQLite and cannot be used, hence, why k was selected above. Attempting to use key will cause the table's creation to fail.

Working with data​

Adding key-value pairs to the table is straightforward. You can INSERT the data with some key and value, and since the type is TEXT, it is rather flexible to handle a UUID or unique identifier (generated outside of Tableland).

INSERT INTO
  key_values
VALUES
  ('key-or-uuid', '{"some_json_key": "some_value"}');

This is exactly the type of repetitive query string that you can easily wrap in an app or use a library to make it even easier. If you want to retrieve the data a specific key, it follows standard SQL read convention.

SELECT
  v
FROM
  key_value
WHERE
  k = 'key-or-uuid';

The resulting query will give you the JSON values.

v
{"some_json_key": "some_value"}

Since k is a primary key, it is guaranteed unique and will return a single value v. Thus, you have key-value data in relational tables. Imagine linking arbitrary JSON data to other structured data via its key, or storing encrypted blobs of data in a TEXT → BLOB key-value-style table. There are an endless number of possibilities, and the relational model with SQL makes everything a lot easier to work with.