Purpose
I built a full-stack application that displays the network latency of randomly generated currency trades, sent between two Docker containers, using Python, NextJS, MongoDB, Docker, Ubuntu, Scapy, and Bootstrap. This application also has the ability to record trades into an AWS EC2 server for later use.
Format & Functions
This project has two modes: normal and recording, these modes are determined based on a “status” flag in MongoDB. This status flag can be altered by clicking the record button in the front end, which is built with NextJS (TypeScript). The frontend contains two NextJS REST APIs, one for getting the 50 most recent trades, and another API used to query the status flag from the database.
The first part of the backend revolves around its “normal” mode. Normal mode enables the tracking of network latency by sending packet payloads between two Ubuntu-based Docker containers. In normal mode, the first “exchange” docker container sends information to a local network over UDP. The second “trader” container reads and sends it to a MongoDB database.
The second part of the project revolves around its “recording” mode. If the system is in recording mode, then the “exchange” docker container sends trades to an AWS EC2 server. The EC2 “recording server” instance has two services: a “system status monitor” service for monitoring the database status flag and enabling the trade logging service, and a “trade logging” service for recording UDP packets into a text file. Furthermore, Monit is used to monitor whether the "trade logging" service is enabled or disabled.
The first part of the backend revolves around its “normal” mode. Normal mode enables the tracking of network latency by sending packet payloads between two Ubuntu-based Docker containers. In normal mode, the first “exchange” docker container sends information to a local network over UDP. The second “trader” container reads and sends it to a MongoDB database.
The second part of the project revolves around its “recording” mode. If the system is in recording mode, then the “exchange” docker container sends trades to an AWS EC2 server. The EC2 “recording server” instance has two services: a “system status monitor” service for monitoring the database status flag and enabling the trade logging service, and a “trade logging” service for recording UDP packets into a text file. Furthermore, Monit is used to monitor whether the "trade logging" service is enabled or disabled.
Experience & Process
I built this project to accomplish a few goals. I wanted to experiment with infrastructure automation, explore Ubuntu configuration/monitoring/alerting, explore tools and concepts related to Linux systems and system administration, and get a better understanding of networking concepts. I decided to learn more about these concepts by combining all of the aforementioned concepts with my knowledge of software engineering to build a hands-on project. While building this tool I overcame many challenges relating to firewall configuration, packet/crafting delivery, and inter/intra-network communication.
Languages & Tools Used
- Frontend: TypeScript/ReactJS/NextJS (ReactJS framework)
- Backend: Python & MongoDB
- Containerization/OS/Servers: Docker, AWS EC2, Ubuntu 22.04
- Python Libraries: Scapy (Packet sniffing and creation) & Pymongo (MongoDB access)
- Ubuntu System Level: Systemd & Monit
Next Steps
- Use Ansible for EC2 instance auto-deployment
- Finish setting up SMTP mail server for Monit notifications from EC2 server
- Set up EC2 monitoring dashboard using Prometheus Node Exporter and Grafana
- Create (or utilize an industry-standard) method to detect and resend dropped UDP packets
- Add packet replay feature, enabling the sending of packets from the EC2 logfile at varying speeds