Synchronous calls between services create tight coupling and lead to cascading failures when downstream systems experience latency.
Using message queues like RabbitMQ isolates high-latency services and improves system resilience by changing from immediate creation to eventual acceptance.
Slicing microservices based on data entities like 'user' or 'account' creates dependencies across teams and slows down feature development.
Reactive systems push data changes through event streams, allowing services to build local read models (projections) and achieve runtime autonomy.
Eventual consistency involves three main challenges—divergence, variance, and latency—which can be managed with ordering, deduplication, and optimistic UIs.
Reactive architectures improve resilience and enable horizontal scaling by creating autonomous services that can build real-time, push-based user interfaces.
Change Data Capture (CDC) streams events directly from a legacy database's transaction log, enabling gradual monolith decomposition without modifying the legacy code.
Sharding partitions data and processing across independent streams, enabling systems to scale horizontally and handle nearly unlimited throughput.
Evolving from a monolith to a sharded, stream-based system adds powerful capabilities but also increases complexity, so always choose the simplest architecture that meets your requirements.
The speaker discusses how micro-frontends align with team autonomy and clarifies that while data can be copied to projections, updates must go through the owning service.
41:06David Leitner
43:27Allard Buijze
28:12Tatiana Chervova
43:13Carola Lilienthal
58:23Luka Samaržija & Ivan Sokol
40:17Luca Mezzalira
45:48Denis Washington & Olli Salonen
43:20Josh Goldberg
.gif?w=240&auto=compress,format)
Jobs that call for the skills explored in this talk.
M&M Software GmbH
Sankt Georgen im Schwarzwald, Germany
CONTIAMO GMBH
Berlin, Germany
Peter Park System GmbH
München, Germany
Speech Processing Solutions
Vienna, Austria