Managed infrastructure for SMBs
Staging and Production Servers for Small Game Studios: Test Builds Before Players See Them
The short answer
A staging server is a copy of your live game server or backend where you test a new build before players ever see it. Production is the environment your players actually connect to. Small studios often run everything on one box because it is cheaper and faster at the start, and that works until a broken update, a bad config change or a database migration takes the live server down on a Friday night. Separating staging from production does not need a big platform team. It needs a second environment that looks like the first, a repeatable way to deploy builds, its own data, and a simple rule for promoting a build from one to the other. This guide covers the setup in ten steps, with a checklist you can reuse for every release.
1. Decide what counts as production
Write down every piece players or customers depend on: game servers, matchmaking or lobby services, login and account APIs, the database, the store or donation page, the website and the Discord bot if it touches accounts. That list is production. Anything on it should only change through a release you have tested. Tools that only your team uses, such as build machines and internal wikis, can live elsewhere and follow lighter rules.
2. Keep staging separate, not just a second folder
Staging should run on its own server, virtual machine or container, with its own network address and hostname such as staging.example.com. A second folder on the production machine shares the same CPU, memory, disk and ports, so a runaway test build can still slow down or crash the live game. Separate machines also stop a staging script from writing to the production database by accident because a path or variable pointed at the wrong place.
3. Make staging look like production
Use the same operating system, the same runtime versions, the same game server version and the same configuration layout. Staging can be smaller in CPU and memory, but it should not be different in kind. Bugs that only appear on a different Linux release or a different Java or .NET version are exactly the ones a staging server is meant to catch. If you are unsure whether to use full virtual machines or containers, KVM vs LXC explains the trade-offs.
4. Deploy the same way every time
Pick one repeatable method to put a build on a server: a script, a container image, a CI pipeline or a panel that pulls a tagged release. Use that same method for staging and production. If staging gets a careful manual copy and production gets a rushed one, you are not really testing the release you ship. Tag each build with a version number so you always know which one is running where.
5. Keep configuration and secrets out of the build
The build should be identical in both environments; only configuration changes. Keep things like database addresses, API keys, store webhooks and payment keys in environment variables or a secrets file per environment, never inside the build itself. Staging should use test keys for payment providers and its own database credentials, so a test purchase or a cleanup script can never touch real money or real player accounts.
6. Give staging its own data
Never point staging at the production database. Use a copy of production that is refreshed on a schedule, with personal details such as email addresses removed or replaced, or a seeded test dataset. A recent copy matters because many bugs only appear with real world sizes, old save formats or odd edge cases that a clean test database will never contain. Restoring that copy into staging also proves your backups actually restore; see how to restore a game server backup.
7. Test the things that break live servers
Before promoting a build, check the boring failures first: the server starts cleanly, players can connect from outside your network, saves load from the previous version, plugins and mods still load, the database migration runs and can be reversed, and the server shuts down without corrupting data. Then run a short playtest with your team or a small trusted group. Playtest readiness covers what to have in place before you invite outside players.
8. Promote with a short release routine
Write the release steps down once and follow them each time: announce a window to players, take a fresh backup of production, deploy the exact build that passed staging, run the migration, check logs and connect as a player, then post that the update is live. Small studios do well with a fixed release day and time when someone is around to watch the first hour, rather than shipping late at night.
9. Always have a way back
Every release needs a rollback plan you have actually tried in staging: keep the previous build on the server, know how to switch back to it and know how to restore the database if a migration goes wrong. Automatic scheduled backups help, but take a manual one right before each release so the restore point is minutes old, not hours. Schedules and auto backups walks through setting those up.
10. Watch production after every change
Monitoring tells you a release went wrong before your players do. At minimum, alert on the game server process stopping, the port no longer answering, disk filling up and memory running out, and keep logs long enough to compare before and after a release. VPS monitoring and alerts covers a simple setup, and securing SSH keeps the servers themselves locked down.
Release checklist
- Production list written down: game servers, backend services, database, store and website
- Staging on its own server or container with its own hostname, not a folder on production
- Same operating system, runtime and game server versions in staging and production
- One deploy method used for both environments, with tagged build versions
- Configuration and secrets kept per environment, test payment keys in staging
- Staging data copied from production on a schedule with personal details removed
- Startup, outside connection, old saves, mods, migration and clean shutdown tested
- Release window announced and a fresh production backup taken before deploying
- Previous build kept and the rollback tested in staging
- Monitoring and logs checked for the first hour after the update goes live
Planning ahead for launch? Read infrastructure before launch and how to grow without overbuying.
Next steps
GlassHosting is a managed service and infrastructure provider: we run your servers on our own network, so you can run your business. If your studio would rather spend its time on the game than on running staging and production servers, read what managed hosting includes and managed vs unmanaged VPS, browse the plans, then tell us what you run or email sales@glasshosting.com.
Host with GlassHosting
Order a plan or browse the knowledgebase for deeper how-tos.