ironSource Reduces Instances Count by 21% with Granulate

Download Case Study

25%

REDUCED CLOUD COSTS

21%

REDUCED INSTANCES COUNT

29%

INCREASE IN THROUGHPUT

2

WEEKS TO RESULTS

AdTech
HQ: Tel Aviv, Israel
Employees: 1,100+

ironSource enables mobile content creators to prosper within the app economy. They support developers as they launch, monetize and scale their apps and games by providing solutions for app discovery, user growth, content monetization, analytics and publishing.

The Challenge

Infrastructure: Scala, Node.js on EC2, EKS on AWS

ironSource is running 100% on AWS, with 2,000 spot instances, and five Kubernetes clusters. They use Scala and Node.js as their main programming languages.

The AdTech company works with 25 billion daily ad requests, a 100 terabytes data pipeline, and online and offline machine learning modules processing 4 billion daily events.

In the real-time bidding industry, maintaining low latency is crucial to win more ad requests and can directly impact the bottom line. So, ironSource strives to maintain low cloud costs while constantly improving CPU and latency.

We saw a quick win in terms of cost reduction, with almost no R&D resources, when every other cost reduction approach required a lot of resources and disrupted the team from moving forward with the business plan.

Gil Shoshan, Sr. Software Developer

The Results

ironSource saw a quick win in terms of cost reduction with almost no R&D efforts required, as compared to other cost reduction approaches which require a lot of resources and distract the team from moving forward with the business plan.

At first, ironSource deployed Granulate’s agent on one of their heaviest servers – the application bidding server.

After one week, they saw a 20% decrease in instances count with no change to latency. Results were immediate, with only one day of DevOps hours needed.

After the successful POC, ironSource deployed the agent on more and more servers, and saw similar results.

With Granulate, ironSource was able to process 29% more throughput and reduce its cloud bill by 25% across their entire online ad serving stack within 2 weeks of deployment.