Engines and proxy tiers

Leave engine and proxy_tier out. We start on the cheapest route and move up only when the site blocks us. Set them only when an answer tells you to.

The routes

A route is one engine on one kind of exit IP. We call each one a rung.

The HTTP engine sends one request that looks like a real Chrome or Firefox on the wire. X-Identity says which. The browser is a real Chrome that runs the page’s JavaScript. Credits and pricing has the price of each route.

The climb

With engine and proxy_tier left out:

1. We fetch over HTTP on a datacenter exit, the cheapest route.
2. When the site blocks us, we try the next route your plan allows.
3. A try that ended in a block we detected costs nothing.

X-Hops lists every try, in order, with its time, as a JSON array: [{"rung":"h:dc","identity":"chrome","ms":412,"outcome":"blocked"},{"rung":"c:dc","identity":"chrome","ms":2900,"outcome":"ok"}]. outcome says how each try ended: ok, blocked, needs_js, or the error code of a try that failed. When an answer is blocked, X-Untried lists the routes we did not try yet.

When to set them yourself

• X-Hint: needs_js. The page builds itself with JavaScript, and HTTP got only the empty shell. The answer is free. Send engine=browser.
• X-Hint: browser_blocked. The site blocks our browser too. The answer is a free 422 blocked.
• The browser on residential IPs. Send engine=browser and proxy_tier=residential. We never pick this route on our own.
• One visitor across requests. Send the same session_id on each. A session uses a residential exit unless you send proxy_tier=datacenter.

The browser takes GET only, with no session_id and no identity. Any of them with engine=browser is a free 400 bad_parameter that names the one to drop.

Ask first

GET /v1/tier?host= lists, under capabilities, the routes that can serve you right now. Check a site first shows how.