This is a time capsule and some of my favorite stories of where I’ve been and who has come with me prior to joining Zynga as Director of Engineering for Kubernetes. Historically, my teams and I were lucky to build games and advise studios (both indie and AAA) starting in-engine client-side and then the backend over 15 years.
I care about investing in people to move architecture and build games. I am attracted to the intersection of security, technology, business, liveops, availability, and observability. If you’re here, I hope you too wake up excited to thrive within that Venn diagram of shared responsibility and infrastructure complexity.
Past Architectures
The below is a reference architecture I created to teach serverless for turn-based games prior to joining AWS to a smaller studio who was evaluating against Photon. At AAA scale I have seen cloud costs drop by almost 2/3s from legacy stacks that were running long-lived EC2 + RDS style infrastructure over 5 years.
AWS Lambda functions (or similarly Azure Functions if you are into that) and DynamoDB are often the cheapest way to build mobile backends while giving owners fine-grained control – The con is teams need to manage all of the above themselves. A lot of this passion to share about serverless was fueled by Ker-Chunk Games‘ own journey in ’15-’20 to investigate if we could move to serverless from a traditional EC2 + RDS stack.
In ’20, I also built a simple serverless demo hosted on S3, API Gateway, Lambda, DynamoDB called prepareforcrits.com which lets you dice roll and feed yourself architecture about being prepared for a DDOS attack. 5 months later, I got “the opportunity” to save a studio’s launch day alongside an amazing team when a 5 AM page showed I had predicted a DDOS on a SPOF.
Unity Build Pipelines with EC2 Mac (2021)
With the launch of EC2 Mac, studios could split the Unity build process in half and only build the Xcode portion on EC2 Mac – this makes mobile builds faster by first auto-scaling over ephemeral instances before hand-off to Xcode to build for iOS. We had to let the world know! Arguably this was where I stepped into Kubernetes for build use cases and containers.
In a life-defining 12-month AWS adventure that ended in a Q4 ’21 award, I was part of a “build farm team” of global SAs led by a visionary GTM specialist (Suvi Tanninen). We released the above on AWS reference architectures (click here for detailed architecture).
Games Streaming with NICE DCV (2021)
In late ’20 through ’21, studios explored streaming for remote workstations, Games QA, and press demos. I built a flight team to test NICE DCV’s QUIC UDP implementation in Q1 against other codecs across multiple software suites & games to prove it was performant on 4K monitors at 60 FPS – We wondered, was this the future of the internet? (Turns out: It was early HTTP/3). We released the above on AWS reference architectures (Click here for details as it’s owned by AWS). 12+ months after our research, IETF published QUIC UDP as proposed standard in RFC 9114 June ’22.
Official AWS Blog: Stream a remote environment with NICE DCV over QUIC UDP for a 4K monitor at 60 FPS (2021)
Triple Co-Authors – Myself, Alec Bryan, Antoine Genereux. Special thanks to Proletariat & Team NICE DCV for pushing streaming and QUIC UDP.
Official AWS Blog: Elevate Game Design with Recommendations for Virtual Goods, Quests, and Game Balance using Amazon Personalize (2021)
Myself as Author, Edited By Kyle Somers. Special Thanks additionally to the Amazon Personalize team.
Official AWS Blog: Generate Custom Game Events from Unity Integrated With the Game Analytics Pipeline (2020)
Myself as Author, Edited by Gena Gizzi. Gena also created an amazing workshop. I built a Unity demo to contribute to her incredible work in-engine integration components.
4 AWS Services in 12 Minutes: Amazon S3, AWS Glue, Athena, & QuickSight
Before I joined AWS I felt it was extremely hard to onboard and build analytic stacks for games if you had more than 3 games sending data to DeltaDNA, Google Analytics, or Unity Analytics and wanted control – I showed how to do it by releasing a Python data generator script to go with it with Dragon data based on “How to Train Your Dragon”
Learn AWS Streaming Analytics for FREE! Dive into Amazon Kinesis Streams, Firehose, RedShift, & S3
Most teams usually extend their streaming analytics to include a data warehouse in order to build complex data analysis applications and gain business insights, so I showed how to do this as well by extending KDS to RedShift.
Past Games
Coming soon™! This section is not nearly as robust as I’d like it to be, but for now here is just a sprinkle of trailers and screenshots that still live on of the 25+ apps/games.
But first three companies who have a special place in my heart for which I helped them get to various launches and/or advised on architecture…
