mem-db

Mem-DB is an in memory database, inspired from AWS Dynamo DB and Redis, written in Go. The objective of this project is simple, we create a database that is able to:

• Store multiple tables, with each tables containing items. An item is simply a record in the table.
• Each item is made up of attributes, which is simply a key-val pair.
• The database operations should prioritize faster reads, and should be able to index on the attributes.

Note This project is my attempt at learning Go by creating something that's always intrigued me. How do database indexes work? More generally, can we build a very simple database similar to DynamoDB?

Getting Started

To get started, simply clone the repository in your local machine, and do:

1. Make sure you have Go installed. Here's a guide on how to install it locally.
2. Start the database server:

   go run ./cmd/mem-db-api

3. Get started by inserting a table:

   curl --location 'localhost:8080/tables' \
   --header 'Content-Type: application/json' \
   --data '{
       "id": "12345",
       "name": "first_table"
   }'

4. Insert items into the table:

   curl --location 'localhost:8080/item' \
   --header 'Content-Type: application/json' \
   --data '{
       "name" : "first_table",
       "attributes": {
           "name": "Farhan",
           "message": "Farhan says hello world!"
       }
   }'

5. Query Table items:

   curl --location 'localhost:8080/items?table_name=first_table'

Roadmap

• Define the database data models and methods.
• Finish up all the basic db CRUD operations.
• Implement proper data integrity.
• Write the data structure and algos for Attribute level indexes. Implement an AVL Tree.
• Integrate the index into GetItems() method.
• Use go routines to spin up the server, and implement proper logging.