Dedicated nodes bought one at a time, in the locations you name. Not a whole network: capacity where your current setup is thin, operated by us.
Reply within 3 business hours. No sales sequence.
A single physical node in a single data centre, reserved for your traffic and run by our engineers. You do not rack it, patch it or monitor it.
The machine carries your traffic and holds your objects. Nothing else shares its disk, its memory or its port.
Storage, port capacity, transit and the data centre itself are chosen for the job the node has to do, then written into the quote before anything is ordered.
You get dedicated capacity without buying, racking or replacing a machine. Failures, firmware and lifecycle stay on our side.
A dedicated node appears next to your existing zones. Cache rules, purge and logs behave exactly as they do on the shared tiers.
Most infrastructure conversations start with a single weak spot, not with a map. One node in the right city often settles it.
Updates, DLC and patches land on a known date and every player pulls the same large files at once. A dedicated node in the busiest region absorbs that hour instead of queueing behind other tenants.
Your numbers are fine everywhere except one market, and the fix is local rather than global. A node goes where the p95 is ugly and the rest of the setup stays untouched.
A build goes out and the whole installed base fetches it within a few hours. Dedicated capacity keeps that spike off the shared pool and off your origin.
If the answer turns out to be several regions at once, with your own hostname and routing, you are looking at a Private CDN Network rather than a node. And if the requirement is a country rather than a data centre, with the node possibly sitting in your own building or inside your ISP, that is Local Cache Servers.
A dedicated node does not ask you to move anything. It sits beside the setup you already have and takes the part of the load that setup handles worst.
Keep the incumbent for everything it does well and point one hostname or one region at a dedicated node. Nothing about the migration is all-or-nothing.
A node placed close to your origin absorbs repeated fetches, so the origin answers once per file per TTL instead of once per edge that missed.
Traffic can fall back to the shared network rather than to your origin. Whether that happens automatically or waits for your call is a configuration choice, agreed in writing during scoping. Either way a single dedicated node is an upgrade on top of a working platform, not a new single point of failure.
We confirm what is available there, what transit it gets and what the node needs to hold.
Monthly figure, term and the date the node is live, all in writing. Nothing is ordered until you have that.
The node goes live while everything you run keeps running, so you compare on real traffic before shifting anything.
Add another location when the numbers justify it, or wind one down at the end of its term. Overflow above a node’s capacity can be allowed to run on the shared tiers, or the node can stay hard-capped, whichever suits your build.
Both are dedicated hardware. The difference is whether you are buying capacity or a network.
You are buying capacity in named places. Start with one node, keep your current delivery, add locations when the numbers say so. Routing and hostnames stay as they are.
You are buying a delivery layer: several nodes, your own hostname and certificate, your own routing and your own regional map. See Private CDN Network.
Four platform classes cover almost every single-node build. Each one is preconfigured rather than assembled to order, which is why a node can be specified and shipped without waiting on parts.
The everyday build. Enough flash for a working set and enough memory to answer most requests without touching disk.
For TLS-heavy traffic and many small objects, where request rate rather than object size is what saturates a node.
Four times the flash of the standard build, for a catalogue that has to stay in cache instead of falling back to origin.
The DPU carries network processing instead of the CPU, so the cores stay free for cache work at 200 gigabit.
Every configuration is specified per build. Cores, memory, NVMe, raw capacity and port speed follow the job the node has to do, and they are written into the quote before anything is ordered. Larger builds and mixed racks exist and are quoted the same way.
A dedicated node is a fixed monthly figure, plus delivery on your agreed tier. The figure follows the location, the hardware the job needs and the term. It is quoted per build and it is not a metered rate.
For reference, 100 TB a month on the public tiers costs $415. Whether a node beats that depends on how concentrated your traffic is in one place. The full ladder sits on Custom Enterprise CDN Infrastructure, alongside Hybrid CDN Solutions and Private Cache Servers.
Tell us the location and what the node has to carry. You get a reply within 3 business hours, with either a quote or an honest answer that the shared tiers already cover you.
He reads every request sent from this page and replies within 3 business hours, including the reply that says the shared tiers already cover you. No sales sequence in between.