MENU
| Feature | Traditional Proxy (Squid, HAProxy) | Scramjet Proxy | |---------|-------------------------------------|----------------| | | Request-based, creates new connection per request | Flow-based, persistent multiplexed channels | | Latency | Milliseconds (1-100 ms) | Microseconds (50-500 µs) | | Protocol support | Fixed (HTTP or SOCKS) | Multi-protocol, auto-detected | | Concurrent connections | Limited by threads/processes (thousands) | Massive (millions of flows via goroutines) | | Data transformation | Requires external modules or Lua scripts | Built-in, zero-copy stream transforms | | Resource usage | High CPU per connection | Low, sub-linear scaling | | Ideal use case | Web browsing, basic scraping | Real-time data feeds, WebSockets, IoT, financial tick data |
Unlike traditional proxies that simply forward requests, Scramjet is an . Its core work is centered on three primary pillars: service worker integration, deep request rewriting, and sandbox isolation. scramjet proxy work
: To hide destination URLs from network monitors, Scramjet uses custom codecs to encode and decode web addresses. | Feature | Traditional Proxy (Squid, HAProxy) |
Instead of parsing individual requests as atomic units, Scramjet Proxy uses . It treats all traffic as a series of data flows (defined by 5-tuple: source IP, source port, dest IP, dest port, protocol). Each flow is assigned a lightweight goroutine or fiber (not a heavy OS thread). This allows: Instead of parsing individual requests as atomic units,
Inside the cockpit, Elias watched the throttle levers slide back incrementally. The roar of the engines shifted pitch. It was the 'scramjet proxy work' in its purest form—complex calculus streamed through a radio antenna, keeping the supersonic combustion from blowing itself apart.