Optimizing application performance and compute cost for AdTech

Presented by

Amir Arama - VP Cloud Operations at Perion, Tom Amsterdam - VP Product at Granulate

About this talk

When there's a direct correlation between application performance and generating revenue, it's no wonder the AdTech industry is one of the most competitive ones, comprised of strict SLAs and uncompromising performance requirements. In this session, we review the use case of Perion, a public-traded AdTech company. Perion leverages big data for its products and receives billions of requests to its infrastructure on a daily basis (which also comes with a cost). Being the innovative and data-driven company they are, Perion's dev and engineering teams have already implemented several methods to improve performance and reduce cost: from application modernization (migration from monolith to K8s-run microservices) to using spot and reserved instances. But that wasn't enough. Perion reached a point where it exhausted all optimization practices within its workloads running in the cloud as well as its engineering resources. So what do you do next? Join Amir Arama - VP Cloud Operations at Perion and Tom Amsterdam - VP Product at Granulate as they'll go over: - The main cost and performance challenges the AdTech industry is facing today - Common optimization methods and practices in the market - Progressive solutions to improve performance and reduce cost - Granulate's autonomous, continuous workload optimization and how it helped Perion reduce compute costs by 52%

Related topics:

More from this channel

Upcoming talks (1)
On-demand talks (10)
Subscribers (746)
Granulate’s technology learns the resource management patterns of your application on cloud or on-prem infrastructure and continuously optimizes it at the kernel and runtime level. Customers can start by activating Granulate on just a single VM, but no matter the size of the cluster they will see results of up to 5x throughput, 40% reduced latency and 60% reduced cloud costs. Performance gains come regardless of other optimization solutions in use, and engineers aren’t required to change the code whatsoever.