How MogileFS Powers Scalable Distributed File Storage: Architecture & Deployment Guide

Introduction

MogileFS is an open‑source distributed file storage system originally developed by Danga Interactive, a LiveJournal subsidiary. The Danga team also created well‑known projects such as Memcached and Perlbal. Today, many companies—including Japan's Youpai, Digg, China's Douban, Yihaodian, Dianping, Sogou, and Anjuke—use MogileFS to manage massive image collections.

Unlike traditional network storage that concentrates data on a single server, a distributed file system spreads data across multiple machines, eliminating the server as a performance bottleneck and enabling easy horizontal scaling.

Implementation Components

Server side : Consists of mogilefsd (the tracker) and mogstored (the storage node). The tracker stores metadata in a database (e.g., MySQL) and monitors nodes, while each storage node listens on port 7500 to accept client file‑store requests. After installation, the mogadm tool registers storage nodes in the tracker database.

Utils : Management tools such as mogadm.

Client API : Provides libraries for PHP, Perl, Python, Java, etc., allowing developers to write client programs for file backup and retrieval.

Features and Principles

Application‑level service; no need for special kernel components.

No single point of failure – the tracker, storage nodes, and MySQL (metadata) can be made highly available.

Transport‑neutral; works over NFS or HTTP.

Simple flat namespace – files are stored directly in a domain without directories, identified by unique keys.

Installation and Configuration

The experiment can be run on two machines, each acting as both a tracker and a storage node while also hosting MySQL. Deploy the tracker on each server, register storage nodes with mogadm, and configure the database accordingly.

Experiment Results

Images (omitted for brevity) show the successful deployment and operation of MogileFS across the test nodes.

Nginx Front‑End Reverse Proxy and Load Balancing

The final step configures Nginx with the nginx‑mogilefs‑module to proxy client requests to MogileFS. The module must be compiled into Nginx. After configuration, files become accessible via a short URL such as http://uri/images/aqua.jpg instead of a long path.

Conclusion

For large‑scale sites with massive image storage needs, MogileFS offers a robust solution. MySQL should be made highly available (e.g., master‑slave or semi‑synchronous replication) to avoid a single point of failure. Nginx reverse‑proxying requires the custom module compiled into Nginx.