With over 250 million active live users a month, that demands a huge and powerful tech stack that can not only handle all the plays but must constantly stay one step ahead to provide a challenging, rewarding, and most importantly, a seamless experience for our players.
Overseeing all this is, as you would imagine, a huge responsibility. So, it’s good that our Chief Technical Officer is Steve Collins, who comes with a vast array of experiences in running tech companies (and a PhD in computer graphics), along with a real love of gaming too.
Steve’s been with King for just over two years now and leads the technology strategy and manages the shared technology team. The shared technology platform includes all the technology that we use to power our live games, the data infrastructure, machine learning, payments infrastructure, and the game engines themselves.
With such an extensive overview and remit, we thought it would be a good opportunity not only to talk about where we are in terms of our tech, but where we’re going, and the big challenges of being a future facing company.
King started using Google Cloud services in 2018. Can you tell us how we’ve progressed since then?
Our games are currently primarily served from our own data centres, and we're moving more and more of our capabilities into the cloud, specifically into Google. We started a number of years ago, so it’s an ongoing process that is definitely speeding up, but not being rushed. We’ve been doing it in a very mature way, where our player experience comes first.
Most of our data infrastructure is now in the cloud. Data infrastructure being how we receive all the event information from the millions and millions of devices that players are playing our games on, and how we create all our internal reporting for our data scientists, our business teams, and our engineering teams. Then with all the data warehouses and all the data science tooling on top of that, we can do some really deep and sophisticated analysis of that data, to understand how our players are playing our games, enabling us to create personalised variations of them.
Is the goal to be completely cloud-based?
Yeah, absolutely, we aim to be a cloud-native organisation in the next few years. That's a huge project, because we've got very, very large games. And a lot of backend technology and services and tools that we need to migrate across. So, we have a big team inside of King doing exactly that right now. They’re working with all the other game teams and all the other teams to achieve that in a very safe and managed way that doesn't impact on our player experience. That's number one for us. You can imagine that's quite a big technological challenge. Again, for us at King, the number one priority is that the game experience is not impacted. Players get to play and enjoy our games without disruptions. We can’t impact in a negative way any of that moment-to-moment experience. And yet under the hood, below the surface, there’s so much going on to make that happen. It’s one of the challenges, but also one of the big rewards of the work – especially when you get great feedback from the players.
So, it’s about preparing and tooling up for the future, without disrupting the present?
Yes, it’s exactly that. At King, we can't stand still, this is a super fast-moving industry. And to remain competitive, and at the top of our game, we have to take advantage of the latest technologies, we have to partner with the top companies around the world. And that's the right thing to do from a business perspective. But I think it's great for everyone on the team, because you get to engage with newer technologies and get to move from older systems into newer systems. But we have to do that in a very pragmatic and balanced way. And you can't, as the expression goes, boil the ocean, and change everything all at once. Otherwise, you just get chaos and uncertainty. So, we're doing this in a very managed way.
But I think for us, it's very exciting because the fields are green, we are moving to the cloud, we're able to take advantage of the very latest state-of -the-art and cloud technology: in streaming and in parallelism, and data workloads, and all the open source solutions that are out there as well. So, it's a great time for King and for all our teams.
Is it fair to say that we don’t just throw things out because they might be considered legacy, and we use the right tools for the right job?
I'd say every company has a mix of the cool new stuff that you get to use, and things that are maybe older, that have been around for longer, where there's no real business requirement to move away from that.
If anything, we probably have a fresher set of technologies, and increasingly so, as we move to the cloud. With some of our older workloads, this gives us an opportunity to rethink some of the vendors we work with and rethink the technologies we take advantage of. It's not like we're a bank, with decades-old pieces of COBOL, living in an old RPG server that sits in a dusty basement, we’re constantly updating. But as you say, we use the right tools for the job. We use Java across our servers, which is old technology at one level, but it's absolutely the right technology for implementing server code and we’ll continue to use that.
Tell us about the platforms and programs you mainly work with.
There’s a lot – how long have you got?
Let's focus on the cloud, because that's where we're moving to.
There’s a lot, even within that. We look at using all the capabilities of our cloud provider, we evaluate very carefully the solutions that they provide and make sure they fit our enormous workload scale. We take full advantage of different compute instances, we take full advantage of things like Kubernetes which is our infrastructure for scaling up quickly and defining configurations of services. And we're big users of BigQuery, which is essentially the database or data warehouse infrastructure. We also use Looker, which is their business intelligence tooling, on top of that and which is used throughout the company.
And then, outside of that, we're using Kafka for streaming. And we’re also taking advantage of some of the Google-specific infrastructure, which is exciting. There’s a very scalable database infrastructure called Spanner, something that we're pretty excited about because of its ability to ingest vast volumes of data. And then to make that data available different availability zones and regions is important to us.
Increasingly we're taking advantage of machine learning at King, so we use AutoML for building machine learning models. And some of the other capabilities of TensorFlow, an end-to-end open source machine learning ecosystem, and others. I mean, there's a long, long list, but these are some of the highlights.