Cloud Native Computing Foundation KubeCon + CloudNativeCon North America 2019 (San Diego), 20 Nov 2019

Previous Meeting

⏯

youtube image

►

From YouTube: Lightning Talk: Open Match - Matchmaking Framework - Robert Bailey, Google

Description

Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io

Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects

Lightning Talk: Open Match - Matchmaking Framework - Robert Bailey, Google

Open Match is an open source game matchmaking framework that simplifies building a scalable and extensible Matchmaker. Open Match is designed to give the game developer full control over how to generate quality matches without having to deal with the challenges of building a scalable online production service. It enables the game developer to re-use the core framework code across games, and just focusing on the rebuilding the matchmaking logic custom to each game. At its core, Open Match comprises of a set of services hosted in a Kubernetes cluster that manage Players, trigger custom matchmaking logic to generate match proposals and provides ability to evaluate these for quality. The framework also provides functionality such as monitoring, alerting, metrics analysis, autoscaling etc.

https://sched.co/Uaf2

A

All right, let's see, can we get a show of hands of people out in the audience who like to play online games all right lots of you excellent. So hopefully, this talk is exciting and relevant. My name is Rob Bailey I work on Google's cloud gaming team. One of our charters is to help the game industry with open source tools to help them build games better and faster, and today what we're gonna talk about is matchmaking. So matchmaking is a really really important part of building an online multiplayer game.

A

You can imagine if you're playing a game and you get if you're really good at a game and you get matched with somebody who's really not very good at a game. It's not a great experience for you and it's a really bad experience for them. You can also imagine if you're trying to get matched for a game, and it takes a really long time. Then you're not gonna, want to play that game right. So there's a whole bunch of attributes need to look at player skills.

A

You need to look at latency to the game servers and need to figure out how many people you need to make a match, and so you have these player populations and you have matching code and basically, this gets rewritten new, every time, you're, building a new game, and so Google decided to team up with unity and start building a framework to build matchmaking services. We wanted this to be an open source framework that way it could be easily extensible, it's very flexible and particularly relevant to this audience.

A

It's written on top of kubernetes certain is a series of micro services where you plug in your matchmaker as a service inside your cluster, and then we leverage kubernetes to make the matchmaking system flexible and highly scalable. What that ends up. Looking like in practice, is you have your game front-end with your core game logic that calls into the matchmaking framework.

A

At the same time, you have a sort of a pool of servers that players can be placed on when they're playing the game and that's simultaneously calling into the matchmaking logic and, as you have players coming and going, you also have server availability, sort of increasing and decreasing, and the framework is sitting in the middle sort of adjudicating between the players that are available splitting them by skill level by latency.

A

You know where they are, where they're located and matching them up with servers that are available, and this is something that you can then reuse across your games as you as you build multiple games in the future and what that looks like. So, if say, if you're building a really small game at that point, really you're not interested in in a really complex, highly scalable matchmaking framework.

A

But you can still leverage open match to really write a really simple matchmaking function, that sort of like a single function in your language of choice and plug that into the framework and be off to the races without having to do a lot of work.

A

But we're open match really shines is, if you have a really large system with hundreds, thousands, millions of players that are all coming in at the same time and the player population is segmented into different groups and matched up very quickly right, and so you might be running hundreds of sort of concurrent matches on the player population.

A

Looking for the best match to make your game work so now we're gonna do something a little bit risky and we're gonna do a live demo during a lightning talk and we're gonna see if we can get this to work. All right looks like my screen got a little bit small.

A

So what we're gonna do is we've got a cluster running.

A

And hopefully we hit excellent all right, so we have a matchmaking function deployed and what we're gonna do is we're gonna start a game front-end, and what this is gonna do is it's gonna write, it's gonna create a front end. It's gonna artificially create a hundred thousand players right, so we're generating a hundred thousand players into our server and saying all these people really want to have a match.

A

So we're gonna make sure that it's up and running.

A

Excellence and then we're going to see that it's finished all right, so we've create a hundred thousand tickets, took a few seconds to do that now. What we're gonna do is we're gonna.

A

There we go we're gonna, do is we're gonna create the actual matchmaker right. So this is gonna, go through those hundred thousand tickets and try to group them into different pools.

A

So, to make sure that it's running.

A

One second excellent all right and we're do the same thing. We're gonna check the logs for this guy and we can see that we generated a hundred thousand matches with a hundred on tickets in just 10 seconds right. So we've got a huge number of players coming in and we're able to match all those players really quickly. What's interesting is, if you look at the logs a little bit higher, you can see that we're actually matching them into different groups right. So some people are playing battle.

A

Royales I'm bill to playing capture the flag, and some people are in sort of our demo profile. So we have lots of different match profiles coming in and they're all being matched simultaneously.

A

If you want to learn more, come, find us on github or on slack or follow us on twitter, and if you want to come chat more, please come find me at the Google booth over the next couple days. Thank you so much.