Talks

How to implement a native Rust engine in a Java application via Project Panama (FFI). Practice and pitfalls.

Why the service layer should return an Entity, not a DTO. Why fetching small collections can lead to million-dollar datasets. Is it worth fetching deep-nested entities. How turning off autocommit can dramatically increase application throughput.

I will tell you about the structure and operation of the main debugging tools in JVM. I will give examples of debugging distributed applications, and also tell you about other non-trivial methods of debugging in JVM.

A modern senior developer working with Spring is simply obliged to understand Kubernetes. In the talk, you will find out what you need to know, what the pitfalls are, what to look for, and how to use deployment tools in Kubernetes.

With the appearance of virtual threads, ThreadLocal is no longer the only solution for context transfer — Java 21 introduced ScopedValue. But when to use what? Which is faster and uses less memory?

I will trace the path of retries from repeated attempts in real life to their implementation in microservice architectures. I'll tell you how retries have evolved in life, programming, and the Java ecosystem. Which strategies and tools help to achieve reliability, and which ones break everything.

I'll tell you how some of the old (compressed oops and compressed class pointers) and new (compact object headers) settings affect actual memory usage, and what to do if your application is approaching the 32 GB limit.

If your “hello world” Spring app needs 512 MB RAM and launches in 10 seconds, you’re not building an app — you’re running a memory leak with REST endpoints.

Using examples, I'll show you which metrics I recommend collecting for an abstract web application (for each component, including APIs, databases, and queues) to make it easier and faster to investigate incidents and various performance issues.

Rich Errors: reinvention of Checked Exceptions. I will tell you the background of the appearance of Rich Errors, what alternatives there are in other ecosystems and why we will never have real Union types.

I will tell and show with examples of real code how we at SberDevices use coroutines and flow to solve non-standard problems. I will consider the most interesting examples of solutions, and at the same time I will analyze the problem of transition between the world of coroutines and the world of legacy callbacks.

We will discuss the issue of the development with Spring in conjunction with JPA, the related performance problem, and also suggest a solution.