Hosting, ancho de banda e ingeniería de redes, escrito por el equipo que opera el AS 25198 — capacidad sin límite, peering, almacenamiento y las decisiones que hay detrás. Los artículos se publican en inglés.
Renting turns hardware into a monthly line item; colocating turns it back into an asset with jobs attached. The technical difference is small, the financial one decides everything. Build both sides over three years — and cost your own engineer's time honestly.
One exit node is easy to specify and tells you nothing about the next fifty. Concurrency, throughput per session and oversubscription set subscribers per node; packet rate rather than cipher choice sets the ceiling; and country coverage wants a tiered fleet, not identical boxes.
No one can quote a player count without knowing the title, but the shape of the limit is knowable. The tick deadline binds first, memory decides instances per box, and bandwidth binds the fleet rather than the instance. Each has arithmetic you can run on your own numbers.
A shared slot and a dedicated node can cost the same per terabyte and behave nothing alike. Contended random IO, fair-use ceilings and shared address reputation are what separate them — along with everything a slot simply cannot run.
A VOD box is two machines pretending to be one: a large cold catalog and a small hot working set. Price them separately. From 2.7 GB per 1080p hour through 96 TB raw becoming 52 TiB usable, to why range requests are the argument for an NVMe tier.
A streaming server does three jobs with three different bottlenecks. Ingest wants path quality, transcode wants cores, egress wants a port. Size them separately, work a real build end to end, and see why the blended ABR bitrate — not the top rung — sets the uplink.
Monthly transfer is the wrong starting point, because a port is a rate. Size from peak concurrency, per-unit bitrate and headroom — then check the cost per gigabit of each tier, because that curve is not monotonic and the cheapest gigabit is not where most people buy.
A virtual machine is a slice of a box; a dedicated server is the box. The useful question is where the slice stops being enough. Four ceilings decide it — single-thread speed, disk latency, packet rate and RAM — and an afternoon of measurement settles it better than a year of guessing.
The machine is the easy part. Address reputation, geolocation and an abuse process that answers decide whether the service works at all — then conntrack, MSS clamping and IPv6 decide whether it scales past one node.
A game server is a simulation loop with a fixed deadline, and no amount of cores makes one thread finish sooner. What to buy for a single instance, what to buy for density, and why jitter beats port size.
The edges take the load until a purge, a deploy or a cold region sends every one of them to the origin at once. Hit-ratio arithmetic, the cache headers that are the whole contract, and which machine to actually build.
The failures in colocation are rarely technical: missing rails, a dual-PSU server on one feed, nobody with the IPMI password. Space, power, network, remote access and paperwork, in the order they cost you a day.