Cryptex

it is a real-time crypto trading engine built using Redis, WebSockets, and a microservices-based API style, its is well monitored logged and oberved using Loki, Promethues, Grafana

real-time crypto trading engine built using Redis, WebSockets, and a microservices-based API system

Demo Link: https://drive.google.com/file/d/1GHW_pp2NaNt9nJgo-UVYER41yEj39kXJ/view?usp=sharing

User Flow

1. Browser → API (/api/v1/order):

   • A client sends a buy/sell order.

   • Example:

     {
     
       "kind": "buy",
     
       "type": "limit",
     
       "price": 12000,
     
       "quantity": 20,
     
       "market": "sol/usdc"
     
     }
     

2. API → Redis Queue:

   • The API pushes the order into a Redis Queue using a key like MESSAGES.

3. Engine (Matching Engine):

   • Subscribes to Redis Queue.

   • Pops the order using the MESSAGES key.

   • Processes the order and publishes the result (fills, executed quantity) using Redis Pub/Sub to the respective clientId.

4. API → Redis Pub/Sub:

   • API subscribes to Redis Pub/Sub using clientId.

   • Receives real-time updates for order execution (ORDER_PLACED, fills, etc.) and returns this data to the browser.

5. WebSocket:

   • The browser maintains a WebSocket connection for real-time updates.

   • The WebSocket service streams back order status, ticker, trade fills, etc.

---

Engine Internals

• Market Streams on Init:

  • On startup, the engine listens to Redis streams like:

    • Trades@SOL/USD

    • DEPTH@SOL/USD

    • Ticker@SOL/USD

• Redis Pub/Sub:

  • Used to distribute updates across API and WebSocket components.

• Redis Queue:

  • Handles order queuing for the engine to consume and process.

---

Price Data Aggregation

• Engine → DB Processor:

  • Pushes raw trade data (Price, Volume, Timestamp) into Redis Lists.

• DB Processor → Timeseries DB:

  • Aggregates raw trades into candlestick data (OHLC - Open, High, Low, Close).

  • Stores structured historical data for charting and analytics.

---

Real-Time Pub/Sub Summary

• Subscribed clients are notified through Redis channels using their clientId.

• All Redis messages are structured and pushed through:

  • Redis Queue → for backend processing.

  • Redis Pub/Sub → for client notifications.