A distributed architecture consists of several microservices, where each microservice is implemented using the Actor model such as Akka and runs on a multicore CPU.
Most of the challenges encountered when designing microservices are solved using concurrent design patterns. However, these patterns usually do not take into account the multicore issues and possible optimizations. This is a hard requirement for real-time mission-critical applications, such as a trading platform.This talk will cover best practices for building Reactive design patterns optimized for multicore. To achieve this, we’ll be using the actor model as implemented in our runtime Simplx: https://github.com/Tredzone/simplx
We will present use-cases related to our experience working with Euronext on the next generation trading platform. We will show how to obtain throughput scalability with the number of cores, while keeping the end-to-end latency stable.
C/C++ and Java knowledge is helpful but not required to enjoy this talk.
SHARE THIS TALK
Tredzone, Head of Research & Innovation
Charly is head of Research and Innovation at Tredzone, a software publisher specialized in Reactive software for mission-critical applications. He has been working for more than 10 years in the parallel computing field for multiple industries, at hardware and software domains. His experience led him to a conclusion: all the technical problems encountered in different scientific fields have a finite patterns of solutions. Charly is the author of several scientific publications. He holds an industrial PhD in computer engineering from Université Paris Sud XI and CEA France.