Performance at Scale

We are teaming up with Riskified and hosting a meetup at DoubleVerify’s offices in Alon Towers, Tel Aviv. Come mingle with and listen to engineers from DoubleVerify and Riskified share their experience working on applications that must perform at scale, serving high volume of requests while meeting strict non-functional requirements such as low latency response and high availability.
---

In this meetup, Alex Cherny, Software Engineering Director at DoubleVerify, will discuss the role of data structures in optimizing high-volume, low-latency systems on the JVM.
Slava Mironov, Staff Engineer at DoubleVerify, will explain the differences between the various garbage collection options and how to pick the right one, and last but not least,
Shai Ungar, Senior Software Engineer at Riskified, will share insights about the process of transforming their system to achieve sub-second real-time analysis.

Agenda:
17:30 – Gathering & Networking - Snacks, Pizza & Beers

18:00 – Data Structures and their impact on high volume, low latency JVM based systems. Alex Cherny, Director, Software Engineering, DoubleVerify
This session explores the role of data structures in optimizing high-volume, low-latency systems on the JVM. We will review a company traffic system and a geo-location use case handling 17 million data points hourly periodic updates. We will examine the advantages and disadvantages of immutable data structures. We will present a comparative performance analysis of various data structures. Lastly, we will discuss scenarios favoring mutable structures for optimal system performance.

18:30 – Picking the right GC to maximize your application performance. Slava Mironov, Staff Software Engineer, DoubleVerify
This talk will build on Alex’s presentation and explain the different GC options and how they affect application performance, specifically for high throughput and low latency services that hold large data structures. We will share how new requirements changed one of our services' performance profile and the steps taken to optimize it.

19:00 – Achieving Sub-Second Real-Time Analysis: A Performance Optimization Journey. Shai Ungar, Senior Software Engineer, Riskified
In the world of e-commerce, a second is crucial as it can lead to cart abandonment and potential revenue loss, while also making fraud prevention harder. Achieving sub-second response times without compromising on extensive machine learning analysis is a significant challenge.
In this session, we will explore the journey of transforming our system to achieve sub-second real-time analysis. We’ll discuss the challenges posed by our initial architecture and how we addressed them by changing our technical stack and utilizing it for offline analysis. Key points will include our approach to setting performance targets, budgeting time for each component, and optimizing dependencies.

Address: Alon Tower#2, 94 Yigal Alon St., Floor 27, Tel Aviv
See you there!