iperamuna / laravel-hypercacheio
High-performance distributed Laravel cache driver using SQLite & HTTP.
Requires
- php: ^8.2
- ext-pdo: *
- guzzlehttp/guzzle: ^7.0
- illuminate/cache: ^10.0|^11.0|^12.0
- illuminate/support: ^10.0|^11.0|^12.0
- laravel/prompts: ^0.1|^0.2|^0.3
Requires (Dev)
- laravel/pint: ^1.27
- orchestra/testbench: ^8.0|^9.0|^10.0
- pestphp/pest: ^2.0|^3.0
- pestphp/pest-plugin-laravel: ^2.0|^3.0
README
Laravel Hyper-Cache-IO is an ultra-fast, distributed cache driver for Laravel applications. By combining L1 in-memory caching with a persistent SQLite WAL backend, it delivers exceptional performance and reliability without the overhead of Redis or Memcached.
Designed for modern PHP environments like FrankenPHP, Swoole, and traditional Nginx/FPM, it features a lightweight internal HTTP API for seamless multi-server synchronization.
⚡ Features
- 🚀 High Performance: Built on SQLite WAL (Write-Ahead Logging) for lightning-fast reads and writes.
- 🏘️ Sharded Architecture: Native 32-shard memory mapping to eliminate lock contention on multi-core systems.
- ⚡ Async Persistence: High-speed background persistence worker for non-blocking writes.
- 🧠 L1 In-Memory Cache: Ephemeral memory caching for instant access during the request lifecycle.
- 🐹 Go Server: High-performance Go-based binary for ultra-low latency and distributed synchronization.
- 🔄 Active-Active HA Mode: Fully synchronized multi-node clusters using binary TCP replication.
- 🛡️ Service Management: Built-in support for running as a systemd (Linux) or launchd (macOS) service.
- 🔒 Distributed Locking: Full support for atomic locks across multiple servers.
- ✅ Modern Compatibility: Fully supports Laravel 10.x, 11.x, and 12.0.
📦 Installation
Install the package via Composer:
composer require iperamuna/laravel-hypercacheio
Run the installation command to configure the package:
php artisan hypercacheio:install
⚙️ Configuration
1. Set the Cache Driver
Update your .env file to use hypercacheio:
CACHE_DRIVER=hypercacheio
2. Configure Server Roles
Hyper-Cache-IO uses a simple Primary/Secondary architecture. You can also run it in standalone mode (Primary only).
Primary Server (Writer)
A single "Primary" node handles all write operations to the database. You can optionally list secondary server URLs so the primary replicates writes to them.
HYPERCACHEIO_SERVER_ROLE=primary HYPERCACHEIO_API_TOKEN=your-secr3t-t0ken-here # Optional: comma-separated list of secondary server URLs for replication. # Invalid URLs are silently ignored. Whitespace around URLs is trimmed. HYPERCACHEIO_SECONDARY_URLS="https://s2.example.com,https://s3.example.com" # Optional: fire-and-forget async replication (default: true) HYPERCACHEIO_ASYNC=true
Secondary Server (Reader)
"Secondary" nodes read from their local copy (synced via shared volume or future replication features) and forward writes to the Primary via HTTP.
HYPERCACHEIO_SERVER_ROLE=secondary HYPERCACHEIO_PRIMARY_URL=https://primary.example.com/api/hypercacheio HYPERCACHEIO_API_TOKEN=your-secr3t-t0ken-here
3. Advanced Configuration
Publish and fine-tune the config file at config/hypercacheio.php:
return [ // Server role: 'primary' or 'secondary' 'role' => env('HYPERCACHEIO_SERVER_ROLE', 'primary'), // Primary server API URL (used by secondary nodes) 'primary_url' => env('HYPERCACHEIO_PRIMARY_URL', 'http://127.0.0.1/api/hypercacheio'), // Comma-separated secondary server URLs for write replication // Invalid URLs are silently discarded to prevent misconfiguration 'secondaries' => HypercacheioSecondaryUrls(env('HYPERCACHEIO_SECONDARY_URLS', ''), ','), // Shared secret for inter-server authentication (X-HyperCacheio-Token header) 'api_token' => env('HYPERCACHEIO_API_TOKEN', 'changeme'), // HTTP timeout in seconds for peer-server requests (recommended: 1–3) 'timeout' => 1, // Fire-and-forget async replication (lower latency, silent failures) 'async_requests' => env('HYPERCACHEIO_ASYNC', true), // SQLite database directory (auto-created by the install command) 'sqlite_path' => storage_path('hypercacheio'), ];
4. Environment Variables Reference
| Variable | Description | Default |
|---|---|---|
CACHE_DRIVER |
Set to hypercacheio to use this driver |
— |
HYPERCACHEIO_SERVER_ROLE |
primary or secondary |
primary |
HYPERCACHEIO_PRIMARY_URL |
Full URL of the primary server's API | http://127.0.0.1/api/hypercacheio |
HYPERCACHEIO_SECONDARY_URLS |
Comma-separated secondary server URLs | (empty) |
HYPERCACHEIO_API_TOKEN |
Shared secret for inter-server auth | changeme |
HYPERCACHEIO_ASYNC |
Enable fire-and-forget replication | true |
HYPERCACHEIO_SERVER_TYPE |
laravel (default) or go |
laravel |
HYPERCACHEIO_GO_DIRECT_SQLITE |
Execute SQLite queries directly in Go | true |
HYPERCACHEIO_GO_HOST |
External/advertised IP of the Go server (used by secondaries) | 127.0.0.1 |
HYPERCACHEIO_GO_LISTEN_HOST |
IP the Go daemon binds to. Use 0.0.0.0 to listen on all interfaces |
0.0.0.0 |
HYPERCACHEIO_GO_PORT |
Port the Go server listens on | 8080 |
HYPERCACHEIO_HA_ENABLED |
Enable Active-Active HA Mode | true |
HYPERCACHEIO_PEER_ADDRS |
Comma-separated replication peers (IP:Port) | (empty) |
HYPERCACHEIO_REPL_PORT |
Port for inter-node binary replication | 7400 |
1. High Performance Architecture (v1.7.0)
The Go server is now architected for Redis-like performance using:
- Sharded Mutexes: 32 cache shards to eliminate lock contention on modern multi-core CPUs.
- Async Persistence: Writes to the SQLite database happen in the background, allowing the HTTP API to respond in sub-microsecond time.
- WAL-Optimized SQLite: Built-in support for WAL mode and synchronous=NORMAL.
📊 Internal Benchmarks (M1 Pro)
| Metric | Performance |
|---|---|
| Parallel GET | 22,000,000+ ops/s |
| Parallel SET | 7,500,000+ ops/s |
| P95 Latency | ~250 nanoseconds |
| TCP Replication | 95,000+ frames/s |
2. Enable Go Server
Update your .env:
HYPERCACHEIO_SERVER_TYPE=go HYPERCACHEIO_GO_DIRECT_SQLITE=true # External IP that secondary servers use to reach this node HYPERCACHEIO_GO_HOST=10.80.3.131 HYPERCACHEIO_GO_PORT=8185 # IP the daemon binds to — 0.0.0.0 (default) listens on all interfaces. HYPERCACHEIO_GO_LISTEN_HOST=0.0.0.0
2. Active-Active HA Mode (TCP Replication)
As of version 1.6.0, Hyper-Cache-IO supports a robust Active-Active HA architecture. Multiple application servers can each run their own local Go cache node, with all nodes synchronizing state in real-time over a dedicated binary TCP protocol.
- Full-Mesh Replication: Every write on one node is instantly broadcast to all configured peers.
- Bootstrap Sync: When a new node joins the cluster, it automatically requests a full state dump from existing peers.
- Zero-Wait Primary: No more bottlenecking on a single "Primary" URL. Your app talks to its local node, and replication happens in the background.
To enable HA Mode, configure your peers in .env:
HYPERCACHEIO_HA_ENABLED=true HYPERCACHEIO_REPL_PORT=7400 # Comma-separated list of peer IP:REPL_PORT addresses HYPERCACHEIO_PEER_ADDRS=10.0.0.2:7400,10.0.0.3:7400
The
hypercacheio:go-servercommand will automatically configure the Go binary to use your application's absolute database path (config('hypercacheio.sqlite_path')) and cache prefix (config('cache.prefix')).
2. Compile & Start
The package includes a full management CLI for the Go daemon:
# 🛠️ Compile the binary for your system (detects macOS/Linux, all architectures) php artisan hypercacheio:go-server compile # 🚀 Start the server as a background process php artisan hypercacheio:go-server start # 📊 Check daemon status (PID / systemd / launchd / process scan) php artisan hypercacheio:go-server status # 🔄 Restart the daemon (Artisan managed) php artisan hypercacheio:go-server restart # 🛑 Stop the daemon php artisan hypercacheio:go-server stop
3. Run as a System Service
Generate service configuration files for your OS, then install and manage them via Artisan:
# Step 1: Generate the service file (systemd for Linux, launchd for macOS) php artisan hypercacheio:go-server make-service # Step 2: Install the service (one-time manual step per OS): # Linux: sudo cp hypercacheio-server.service /etc/systemd/system/ # macOS: cp iperamuna.hypercacheio.server.plist ~/Library/LaunchAgents/ # Step 3: Manage via Artisan php artisan hypercacheio:go-server service:start # Load and start the service php artisan hypercacheio:go-server service:restart # Cycle the service php artisan hypercacheio:go-server service:stop # Stop the service php artisan hypercacheio:go-server service:status # View service status (systemd/launchd output) php artisan hypercacheio:go-server service:remove # Disable and remove the service
🔌 Connectivity Check
To verify that your server can communicate with the configured Primary/Secondary nodes, run the built-in connectivity check command:
php artisan hypercacheio:connectivity-check
This command:
- Identifies server type (LARAVEL or GO) with host and port
- Pings the local node first via
127.0.0.1(falls back to configured host) - Tests Ping, Add, Get, Put, Delete, Lock against all configured endpoints
- Reports OS-specific firewall advice (ufw, firewalld, socketfilterfw) when a connection fails
🛠️ Usage
Use the standard Laravel Cache Facade. No new syntax to learn!
use Illuminate\Support\Facades\Cache; // ✅ Store Data // Automatically handles L1 memory + SQLite persistence + Primary sync Cache::put('user_preference:1', ['theme' => 'dark'], 600); // ✅ Retrieve Data // Checks L1 memory first, then SQLite $prefs = Cache::get('user_preference:1'); // ✅ Atomic Addition // Only adds if key doesn't exist (concurrency safe) Cache::add('job_lock:123', 'processing', 60); // ✅ Atomic Locking // Distributed locks work across all servers $lock = Cache::lock('processing-job', 10); if ($lock->get()) { // Critical section... $lock->release(); }
🔌 Internal API
The package exposes a lightweight internal API for node synchronization. Each endpoint is secured via X-Hypercacheio-Token.
| Method | Endpoint | Description |
|---|---|---|
GET |
/api/hypercacheio/cache/{key} |
Fetch a cached item |
POST |
/api/hypercacheio/cache/{key} |
Upsert (Create/Update) an item |
POST |
/api/hypercacheio/add/{key} |
Atomic "Add" operation |
DELETE |
/api/hypercacheio/cache/{key} |
Remove an item |
POST |
/api/hypercacheio/lock/{key} |
Acquire an atomic lock |
DELETE |
/api/hypercacheio/lock/{key} |
Release an atomic lock |
✅ Testing
You can run the full test suite (Unit & Integration) using Pest:
vendor/bin/pest laravel-hypercacheio/tests
📄 License
The MIT License (MIT). Please see License File for more information.
❤️ Credits
- Developed with ❤️ by Indunil Peramuna