Synchronous Multiplayer Using Azure Kubernetes Service (AKS)

• 3/14/2019
• 4 minutes to read

You can choose to manage containerized dedicated game servers using the Kubernetes orchestrator on Azure with the managed Azure Kubernetes Service (AKS).

This article will describe the architecture used in this sample on GitHub. Keep in mind that the code from this reference architecture is only an example for guidance and there may be places to optimize the code before using in a production environment.

Architecture diagram

Relevant services

• Azure Traffic Manager - Cconnects the player to the most appropriate regional zone based on latency.
• Azure Kubernetes Service - Simplifies the deployment and operations of Kubernetes.
• Azure Container Registry - Allows storing images for all types of container deployments.

Architecture considerations

Custom resource definitions (CRDs)

In this reference architecture, Kubernetes is extended by using Custom Resource Definition (CRDs) objects. These objects will be used to represent the dedicated game server entities.

Specifically, there are 2 core entities, which are represented by 2 respective CRDs:

• DedicatedGameServer: represents the multiplayer game server itself. Each DedicatedGameServer has a single corresponding child Pod which will run the container image with the game server executable.

• DedicatedGameServerCollection: represents a collection/set of related DedicatedGameServers that will run the same Pod template and can be scaled in/out within the collection (i.e. add or remove more instances of them).

  DedicatedGameServers that are members of the same DedicatedGameServerCollection have a lot of similarities in their execution environment, e.g. all of them could launch the same multiplayer map or the same type of game. So, you could have one collection for a "Capture the flag" mode of your game and another collection for a "Conquest" mode. Or, a collection for players playing on map "X" and a collection for players playing on map "Y".

Components

This reference architecture contains 2 main components, both of which are created as a single instance Kubernetes Deployment:

1. API Server components

   This is our project's API server (that has nothing to do with Kubernetes API Server). It contains two sub-components:

   1. API Server subcomponent

      It provides REST APIs that can be called by either the game server or an external scheduling system (e.g. a lobby service or a matchmaker).

   2. Webhook subcomponent

      A Kubernetes admission webhook, it validates and modifies requests about our CRDs to the Kubernetes API Server.

2. Controllers

   Kubernetes controllers are objects that are active reconciliation processes. In simple words, this means that a controller watches an object (or a set of objects) for its desired state as well as its actual state. It actively compares them and makes every effort to bring the actual state to look like the desired state.

   This project contains a set of controllers, each one carrying the task to reconcile a specific set of objects. All controllers are based on the official Kubernetes sample controller and the respective documentation here. Controllers in this project were made in order to reconcile our Custom Resource Definition (CRD) objects, i.e. the DedicatedGameServerCollections and the DedicatedGameServers.

   1. DedicatedGameServerCollectionController

      It has the duty of handling the DedicatedGameServer objects of a DedicatedGameServerCollection.

   2. DedicatedGameServerController

      It has the duty of handling the Pods of a DedicatedGameServer object.

   3. DGSActivePlayersAutoScalerController

      It is optionally started (via a command line argument on the controller) and is responsible for Pod Autoscaling on every DedicatedGameServerCollection that opts into the pod autoscaling mechanism.

These components are part of their own namespace (named dgs-system in the sample), separated from the namespace of the DedicatedGameServers. In a big cluster scenario, you could have a namespace per list of DedicatedGameServerCollections.

Public IPs

Game clients should be able to connect directly to a Dedicated Game Server, so all Nodes should have a Public IP. Azure Kubernetes Services does not provide a public IP by default, but you can use this GitHub project to enable that.

Deployment template

Follow these instructions to create the Azure Kubernetes Cluster.

Security considerations

All API methods are protected via an access code, represented as string and kept in a Kubernetes Secret called apiaccesscode. This is created during the project's installation and should be passed in all method calls code GET parameter. The only method that does not require authentication by default is the /running method. This, however, can be changed in the API Server process command line arguments.

Optimization considerations

For smaller clusters, you can simplify by using a single Pod (the one containing the API Server subcomponent, Webhook subcomponent and controllers) for all the Nodes. In a Kubernetes cluster, every Pod can talk to other Pods in all other Nodes by design.

Additional resources and samples

Official Kubernetes documentation

Pricing

If you don't have an Azure subscription, create a free account to get started with 12 months of free services. You're not charged for services included for free with Azure free account, unless you exceed the limits of these services. Learn how to check usage through the Azure Portal or through the usage file.

You are responsible for the cost of the Azure services used while running these reference architectures. The total amount will vary based on usage. See the pricing webpages for each of the services that were used in the reference architecture:

• Azure Traffic Manager
• Azure Kubernetes Service
• Azure Container Registry

You can also use the Azure pricing calculator to configure and estimate the costs for the Azure services that you are planning to use.