Colocation
Colocation is placing hardware or workloads side by side in shared infrastructure. In a data center, it means renting rack space, power and connectivity for servers you own. For game servers it more often means many matches, from many studios, sharing the same hosts, each isolated in its own container. Isolation is what makes it work: what each tenant can use, and what it cannot reach.
Also called
colo
,
What Does Colocation Mean for Game Servers?
The word has two meanings, and a game developer searching it usually meets the first one.
Data-center colocation. A company buys its own servers and rents rack space, power, cooling and network connectivity in someone else's facility. The building is shared; the hardware is yours.
Workload co-location. Several programs share one physical host. For game servers this is the common case: many matches, often from many studios, run side by side on the same machine, each in its own container.
Both are co-tenancy, and almost every game server runs with some. A server in a colocation cage shares the building, the power and the network. A cloud server shares a physical host behind a hypervisor. A container shares its host with other containers. What matters is not whether something is shared, but what each tenant is kept from.
Is It Safe to Share a Host With Other Studios' Servers?
Yes, because each container runs inside a budget the host enforces. When a game server starts, it is given an allowance of CPU and memory, and the host's kernel holds it to that allowance through control groups (Docker documentation).
What happens at the edge of the budget stays inside the container:
CPU. A server that tries to use more than its share is held to its limit. A busy match next door cannot take your server's CPU, and yours cannot take theirs.
Memory. A server that goes over its memory limit is stopped. It ends its own match and nobody else's.
That is why a noisy neighbour does not reach a match on a container host: the neighbour's load is capped at the neighbour's budget. Each container also gets its own view of processes, files and network, so one tenant's server does not see another's.
The budget worth planning for is your own. Measure peak memory at your target player count, in a full match, and set the limit above it. A limit sized for the average is the one most likely to stop a match.
Why Does Sharing Make Game Servers Cheaper?
A host you own or rent alone is paid for whether it runs matches or not, and it only exists where you put it. Shared capacity spreads both costs across every game using it: each studio pays for the minutes its servers run, and the cost of keeping a location open is split among many tenants.
That is what lets a small game reach players on several continents. Edgegap's Edge Cloud runs game servers in 615+ locations across 17+ infrastructure providers, billed at $0.00115 per vCPU per minute (platform data, 18 September 2026). Servers can also be smaller than one vCPU: Bloody Bastards uses "fractional vCPU sizing" to run "smaller, right-sized dedicated servers" (case study, Tibith Ltd, 2026).
Data-center colocation works the other way. The hardware is yours, so each location you add is another set of servers bought for your own peak.
Colocation vs Bare Metal vs Cloud: Should a Studio Own Its Hardware?
Owning servers in rented racks can give the lowest cost per busy hour, because no provider margin sits on the hardware. Three things usually decide whether that holds for a game:
Utilization. The hardware costs the same whether it runs matches or not, for its whole life. The busy-hours arithmetic on bare metal server applies, with a longer commitment.
Speed. Adding capacity means buying, shipping and racking servers, which typically takes weeks. A launch bigger than planned needs somewhere else to go.
Sites. Each facility is its own contract, hardware and on-site support. Reaching players on several continents means repeating that for every site.
Colocation tends to fit large, steady fleets run by a team that already operates infrastructure. Rented bare metal keeps much of the economics without the hardware lifecycle, and hybrid orchestration adds cloud capacity for scaling when needed, wider coverage, and the cases where performance decides whether every player in a game gets a good, fair online experience.
A word from our sponsor (ourselves!)
A region is a guess about where your players will be. Edgegap's Edge Cloud is a distributed, multi-cloud network of 615+ locations across 17+ providers, available on demand. Each server launches at the best available location on the network when the match starts, not at the nearest of a handful of regions.
Edgegap's Take (just our opinion, take it with a grain of salt!)
Share the Host, Not the Risk
Studios sometimes rent whole racks or whole machines to keep neighbours away. Containers already do that: every server runs inside its own CPU and memory budget, and a server that exceeds it only affects its own match.
What sharing changes is the bill. A shared host is paid for by the minute, by every game on it, in every location the network reaches. Owning hardware to avoid co-tenancy buys back fixed costs and fewer locations, to solve a problem isolation has already solved.
So we start from the match: size its server to what it needs, run it on shared capacity near its players, and keep owned or reserved hardware for the load that never drops.
,
