Multiplayer I – Introduction

September 2012

This post is a draft. Content may be incomplete or missing.

TODO

We’re currently working on a WebGL port of the nullZERO prototype from CS195-U. Could make sense to revamp the series to use the js port as the example code.

Ever wanted to write a multiplayer video game?

Writing a multiplayer game that works well over a LAN isn’t too hard. Packets moving through a LAN tend to get from A to B quickly. It’s usually enough to replicate the state of your game world between two networked machines.

Writing a multiplayer game that works well over the Internet is a different story. Packet loss is common. Latency is high, and changes without warning. Some players will inevitably tamper with the network to cheat. Dealing with these conditions requires a bit more finesse than simply replicating state.

This series of blog posts is a nuts-and-bolts introduction to building an Internet-ready first-person shooter. The game will perform well over broadband connections, and we’ll take some steps to mitigate cheating.

These posts are based on what I learned while building online multiplayer for nullZERO. I drew heavily from publishings about Valve’s Source Engine and id Software’s QuakeWorld.

Audience

This series is aimed at readers who have experience developing video games. If you can build a rudimentary FPS without relying heavily on sample code, these posts are for you!

Readers should be able to read C++. The example code for each post, as well as the inline code snippets, are written in C++. That said, you can apply what you learn in any language you desire.

Game networking is shenanigans. You should be prepared to get stuck a couple times while following the tutorial. When you do, take a break and come back later!

Outside Reading

Before starting this tutorial series, I highly recommend you read Glenn Fiedler’s Networking for Game Programmers series. In particular, I won’t be covering packet transport layers because Glenn does it so well. He explains client/server gameplay at a high level; this series covers the same in greater technical detail.

The following is a list of other documents I consulted when developing nullZERO. You may also find them useful:

Client-Server in a Nutshell

TODO diagram

We’ll build our game using a client-server architecture. I’ll assume you’re familiar with client/server systems in general. For games, a client-server system usually means:

  • The server tracks the state of the game
  • The server sends the state to the clients
  • The clients locally render the server state to the player’s screen
  • The clients collect player input
  • The clients send the player input to the server
  • The server uses the input to affect the simulation

Posts further down the line will blur this division of labor, but this description will suffice for now.

Terminology

Network programmers worry about time a lot. Each client has a separate timeline of the game’s events, as does the server. As we’ll see later, these timelines can sometimes diverge. To avoid ambiguity, these posts will use the following terms to mean specific things:

  • A simulation is a series of events that occur over time. Time here doesn’t necessarily correspond to the real world: for example, we could choose to run the simulation at half speed. Then 1 second of simultation time would elapse for every 2 seconds of real-world time.

  • Virtual time (or v-time) will always refer to the timeline of a simulation. 1 second of v-time elapsed in our previous example.

  • Real time (or clock time) will always refer to the time elapsed in the real world. 2 seconds of clock time elapsed in our previous example.

  • Client time refers to the v-time of the client’s local version of the simulation.

  • Server time refers to the v-time of the server’s local version of the simulation. In many cases, server time is identical to real time.

If you ever find a place in this series where it’s not clear whether the time being discussed is virtual or real, please file a bug against the blog post!

Example Code

Download: Example Code I

Each post in this series has attached sample code. It’s written in C++ using the cross-platform Qt Framework for graphics, windowing, networking, and the like.

The example code for this post is a game with basic FPS gameplay. We’ll be expanding this into a multiplayer game.

One of the most common beginner mistakes in networking is to decide to build your game first and add multiplayer later. Although you should usually leave networking out of your initial prototypes, ‘adding’ multiplayer usually means rewriting your core game logic. If you want your game to be multiplayer, don’t put it off until the end!

In addition to the FPS skeleton, the example code also implements a UDP protocol like the one Fielder describes in Networking for Game Programmers.

Onward

Ready to start? Head on over to Multiplayer II: Client / Server Architecture!