RADisk Storage Adapter

RADisk is GUN’s default storage adapter for Node.js environments. It stores data as files on the file system using an efficient Radix tree structure for fast queries and range lookups.

Installation

RADisk is included with GUN and automatically enabled in Node.js environments:

const Gun = require('gun');
require('gun/lib/radisk');  // Explicitly load RADisk

const gun = Gun({
  file: 'radata'  // Directory for data files
});

Configuration Options

RADisk accepts the following configuration options:

const Radisk = require('gun/lib/radisk');

const rad = Radisk({
  // Required: storage interface
  store: {
    get: function(file, callback) { },
    put: function(file, data, callback) { },
    list: function(callback) { }  // optional
  },
  
  // Optional configuration
  file: 'radata',              // Base directory name
  log: console.log,            // Logging function
  max: 300000000 * 0.3,        // Max file size (90MB default)
  until: 250,                  // Batch delay in milliseconds
  wait: 250,                   // Alias for 'until'
  batch: 10000,                // Max items per batch
  chunk: 1024 * 1024,          // Chunk size (1MB default)
  jsonify: true,               // Use JSON format (default)
  
  // Comparison function for updates
  compare: function(oldData, newData, key, file) {
    // Return newData to save, undefined to skip
    return newData;
  }
});

Key Options

`file` (string)

Base directory or file name for storing data. Default: 'radata'

const gun = Gun({ file: 'myapp-data' });

`until` / `wait` (number)

Time in milliseconds to wait before writing batched data to disk. This allows multiple writes to be grouped together. Default: 250

const gun = Gun({ 
  file: 'radata',
  until: 500  // Wait 500ms before writing
});

`batch` (number)

Maximum number of write operations to batch before forcing a disk write. Default: 10000

`chunk` (number)

Maximum size (in bytes) for a single data file before it’s automatically split. Default: 1048576 (1MB)

const gun = Gun({ 
  file: 'radata',
  chunk: 1024 * 1024 * 10  // 10MB chunks
});

`max` (number)

Maximum size for individual data values. Default: 90000000 (90MB)

`jsonify` (boolean)

Whether to use JSON format for storage. When true, data is stored as JSON objects. When false, uses the legacy RAD encoding format. Default: true

How RADisk Works

File Organization

RADisk creates a directory structure to store your data:

radata/
├── %1C                  # Directory index file
├── %21                  # Data file for keys starting with '!'
├── user%1Balice         # Data file for 'user/alice' and similar keys
├── user%1Bbob           # Data file for 'user/bob' and similar keys
└── ...                  # More data files as needed

File names are URL-encoded to handle special characters:

%1C - Character code 28 (directory file)
%1B - The escape character separating soul and key
%21 - Exclamation mark (’!’)

Radix Tree Storage

RADisk uses a Radix tree to efficiently organize keys:

// Example: Storing user data
gun.get('user/alice').put({ name: 'Alice', age: 30 });
gun.get('user/bob').put({ name: 'Bob', age: 25 });

// Internally stored in Radix tree:
{
  "user": {
    "/alice/name": { ":":"Alice", ">":1234567890 },
    "/alice/age": { ":":30, ">":1234567891 },
    "/bob/name": { ":":"Bob", ">":1234567892 },
    "/bob/age": { ":":25, ">":1234567893 }
  }
}

Automatic File Splitting

When a data file exceeds the chunk size, RADisk automatically splits it:

Creates a new file for the second half of the keys
Updates the original file to contain only the first half
Updates the directory index to track both files

This keeps file sizes manageable and improves read performance.

Write Batching

RADisk batches writes to improve performance:

// These three writes are batched together
gun.get('user/alice').put({ name: 'Alice' });
gun.get('user/bob').put({ name: 'Bob' });
gun.get('user/carol').put({ name: 'Carol' });

// After 250ms (default), all three writes go to disk in one operation

Benefits:

Reduces disk I/O operations
Improves write performance
Reduces risk of corruption from partial writes

Storage Interface

RADisk requires a storage interface with get and put methods:

const fs = require('fs');
const path = require('path');

const store = {
  // Read a file
  get: function(file, callback) {
    fs.readFile(path.join('radata', file), 'utf8', function(err, data) {
      if (err && err.code === 'ENOENT') {
        // File doesn't exist - this is not an error
        return callback(null, undefined);
      }
      callback(err, data);
    });
  },
  
  // Write a file
  put: function(file, data, callback) {
    fs.writeFile(path.join('radata', file), data, function(err) {
      callback(err, err ? null : 1);
    });
  },
  
  // List all files (optional, but recommended)
  list: function(callback) {
    fs.readdir('radata', function(err, files) {
      if (err) return callback();
      files.forEach(callback);
      callback();  // Signal end of list
    });
  }
};

const rad = Radisk({ file: 'radata', store: store });

Read and Write Operations

Writing Data

// Write a key-value pair
rad('user/alice', { ':': 'Alice', '>': Date.now() }, function(err, ok) {
  if (err) {
    console.error('Write failed:', err);
  } else {
    console.log('Write successful');
  }
});

Reading Data

// Read by exact key
rad('user/alice', function(err, data) {
  if (err) {
    console.error('Read failed:', err);
  } else {
    console.log('Data:', data);
    // { ':': 'Alice', '>': 1234567890 }
  }
});

// Range query
rad('user/', function(err, data) {
  console.log('All user keys:', data);
}, {
  start: 'user/a',
  end: 'user/z'
});

Performance Tuning

High-Throughput Scenarios

For applications with high write volume:

const gun = Gun({
  file: 'radata',
  until: 100,              // Write more frequently
  batch: 50000,            // Allow larger batches
  chunk: 1024 * 1024 * 5   // 5MB chunks
});

Large Data Sets

For applications storing large amounts of data:

const gun = Gun({
  file: 'radata',
  chunk: 1024 * 1024 * 10,  // 10MB chunks
  max: 500000000 * 0.3      // Allow larger individual files
});

Low Latency Requirements

For applications requiring fast writes:

const gun = Gun({
  file: 'radata',
  until: 50,     // Write after 50ms
  batch: 1000    // Smaller batches
});

Data Format

JSON Format (Default)

With jsonify: true, data is stored as JSON:

{
  "user/alice/name": {
    ":":"Alice",
    ">":1234567890
  },
  "user/alice/age": {
    ":":30,
    ">":1234567891
  }
}

RAD Format (Legacy)

With jsonify: false, data uses the RAD encoding:

!#"user"!#"alice"!#"name":"Alice"
!#"user"!#"alice"!#"age":+30!

The RAD format uses special encoding:

! - Field separator (ASCII 31)
# - Prefix length indicator
" - String value
+ - Number value
: - Value separator

File System Considerations

Directory Structure

Ensure the storage directory exists and has proper permissions:

const fs = require('fs');

// Create directory if it doesn't exist
if (!fs.existsSync('radata')) {
  fs.mkdirSync('radata', { recursive: true });
}

Backup and Recovery

To backup RADisk data:

# Simple backup
tar -czf backup.tar.gz radata/

# Incremental backup with rsync
rsync -av radata/ backup/radata/

To restore:

# Extract backup
tar -xzf backup.tar.gz

Cleaning Up

RADisk automatically manages file lifecycle, but you can manually clean up if needed:

const fs = require('fs');
const path = require('path');

// Remove all data files
fs.readdir('radata', (err, files) => {
  if (err) return console.error(err);
  files.forEach(file => {
    fs.unlink(path.join('radata', file), err => {
      if (err) console.error(err);
    });
  });
});

Troubleshooting

Error: “Radisk needs `opt.store` interface”

You must provide a storage interface:

const Radisk = require('gun/lib/radisk');

const rad = Radisk({
  store: {
    get: function(file, cb) { /* ... */ },
    put: function(file, data, cb) { /* ... */ }
  }
});

Error: “Data too big!”

Increase the max option:

const gun = Gun({
  file: 'radata',
  max: 1000000000 * 0.3  // Allow larger values
});

Corrupt File Recovery

RADisk automatically detects and removes corrupt files:

// RADisk will log corrupt files and attempt recovery
// Check console output for:
// "File 'filename' does not have root radix!"

Best Practices

Use appropriate chunk sizes: Smaller chunks for many small writes, larger chunks for fewer large writes
Enable batching: Keep until at 250ms or higher for better performance
Backup regularly: Use file system snapshots or regular backups
Monitor disk space: RADisk can grow based on your data
Use JSON format: The default JSON format is more debuggable than RAD format

Next Steps

Rindexed Adapter

Browser storage with IndexedDB

S3 Adapter

Cloud storage with Amazon S3

Custom Adapters

Build your own storage adapter

Performance

Optimize GUN performance

​RADisk Storage Adapter

​Installation

​Configuration Options

​Key Options

​file (string)

​until / wait (number)

​batch (number)

​chunk (number)

​max (number)

​jsonify (boolean)

​How RADisk Works

​File Organization

​Radix Tree Storage

​Automatic File Splitting

​Write Batching

​Storage Interface

​Read and Write Operations

​Writing Data

​Reading Data

​Performance Tuning

​High-Throughput Scenarios

​Large Data Sets

​Low Latency Requirements

​Data Format

​JSON Format (Default)

​RAD Format (Legacy)

​File System Considerations

​Directory Structure

​Backup and Recovery

​Cleaning Up

​Troubleshooting

​Error: “Radisk needs opt.store interface”

​Error: “Data too big!”

​Corrupt File Recovery

​Best Practices

​Next Steps