Skip to main content

GumGum Improves Performance and Lowers Costs by Moving to ScyllaDB

Published Date

Keith Sader, Engineering Manager of the Ad Server (Demand) team, explains how Gumgum moved to ScyllaDB Cloud to deliver AI-driven ads with better performance and lower costs, while virtually eliminating operational overhead.

Transcript

Hi I'm Keith Sader. Today I'm going to walk you through how we use ScyllaDB at GumGum to serve really relevant ads to users.

We originally implemented our Tally backend in Cassandra. Our Tally Cassandra cluster was 51 instances around the globe. Just the basic cost of that alone was almost $200,000, not including staff time, and our staff time was pretty busy on that. We're not really a Cassandra focused shop. Cassandra is fantastic if you can throw a dev team at it. And we're not. We're making an ad server with computer vision, which is a different problem almost entirely.

What ScyllaDB via Tally does is help us count all the impressions, all the views, all the clicks, all the revenue, all the video streams and play percentages along the way in the course of serving advertisements.

Cassandra wasn't really performing under high loads. And we're not a Cassandra engineering place. So we sort of struggled with that. We thought about replacing this and we went through a bunch of different vendor selections. Originally, we looked at Data Stax; great, a little pricy. We looked at DynamoDB; not really right for us in terms of cost of performance. Plus, we'd have to change the entire Tally back end. Redis? We're pretty read heavy, and we need some consistency in our writes, so Redis wasn't quite it 

ScyllaDB was a drop in replacement. It was also a managed system, so we could really lay off a lot of the operational items to someone else that wasn't us, and we could focus on making relevant contextual ads and serving them.

We really had a lower amount of overhead and a higher throughput from putting ScyllaDB in there. So from 51 instances we went to 30 of ScyllaDB across the globe. Our licensing cost, lower $100,000. Our cluster cost, little over $100,000. But our total annual costs with support was $212,528 This is with the ScyllaDB back end, and they're great. We've got a live Slack channel with them. They're responsive to our emails and queries. So to basically take 20% of two peoples' staff time, an entire load of our off of our system. This was the big win for us. For just 13 grand, got a much better and more performant data system around the globe.