Online Systems

We work out what your backend and publishing stack has to do before anybody writes it. Which platforms you are launching on and what each one demands, how player accounts and purchases carry across them, and what the matchmaker needs to know about a player to put them in a good game. We size the server architecture against the player numbers you are planning for and tell you where it will get expensive.

We build the platform: the backend services, the publishing tools, the live configuration and the features your game needs to roll out globally. We build the matchmaker to your rules, and the systems that scale your servers up and down. Where something already exists, we take it over, fix it or integrate it with the rest. All of it is written to your design, documented, and yours to fork and modify.

We load test the platform against traffic shaped like a real launch and fix what breaks. The matchmaker gets tested against realistic player populations, because one that works with a hundred testers can still fall over with a hundred thousand players. We prove the scaling systems bring servers up fast enough and take them back down without leaving capacity running. We set the latency targets and check the game holds them.

We run the platform through a full rehearsal. Playtests big enough to stress the real services, servers left under load for days to find the slow leaks, and the matchmaker running against launch-sized queues. We write and test the rollback plan for getting the backend back to a known state. We agree the numbers that trigger action on launch day, so nobody is deciding in the moment.

Letting people into the game is the part that fails most visibly, so the matchmaker and the queues get watched live from the first minute. We keep headroom in the services for a launch bigger than forecast and adjust the scaling as the real curve appears. Where the backend needs a change during launch week, we make it and roll it out without taking the game down.

We tune the matchmaker against how your players actually behave, which is never what the design assumed. Queue times get cut where they are losing people. The scaling rules get adjusted to the real player curve, so you stop paying for servers nobody is using. We work through the platform problems the live player base surfaces, and add the publishing features the roadmap needs next.

We keep the platform current as the console holders change what they require. New toolkit versions, expiring certificates and retired third-party services get handled before they become outages. We tune the server scaling as the player base settles into its long-term shape. Where the backend has to move to new infrastructure, we do the migration without taking the game offline.