Real-Time Event Distribution with Data Mesh
Inheriting software in the banking sector can be challenging. Perhaps the only thing harder is inheriting software built by a committee of banks. How do you keep it running, while improving it, refactoring it, and planning a bigger future for it? In this episode, Jean-Francois Garet (Technical Architect, Symphony) shares his experience at Symphony as he helps it evolve from an inherited, monolithic, single-tenant architecture to an event mesh for seamless event-streaming microservices. He talks about the journey they’ve taken so far, and the foundations they’ve laid for a modern data mesh.Symphony is the leading markets’ infrastructure and technology platform, which provides a full communication stack (chat, voice and video meetings, file and screen sharing) for the financial industry. Jean-Francois shares that its initial system was inherited from one of the founding institutions—and features the highest level of security to ensure confidentiality of business conversations, coupled with compliance with regulations covering financial transactions. However, its stacks are monolithic and single tenant. To modernize Symphony's architecture for real-time data, Jean-Francois and team have been exploring various approaches over the last four years. They started breaking down the monolith into microservices, and also made a move towards multitenancy by setting up an event mesh. However, they experienced a mix of success and failure in both attempts. To continue the evolution of the system, while maintaining business deliveries, the team started to focus on event streaming for asynchronous communications, as well as connecting the microservices for real-time data exchange. As they had prior Apache Kafka® usage in the company, the team decided to go with managed Kafka on the cloud as their streaming platform. The team has a set of principles in mind for the development of their event-streaming functionality: Isolate product domainsReach eventual consistency with event streamingClear contracts for the event streams, for both producers and consumers Multiregion and global data sharingJean-Francois shares that data mesh is ultimately what they are hoping to achieve with their platform—to provide governance around data and make data available as a product for self service. As of now, though, their focus is achieving real-time event streams with event mesh. EPISODE LINKSThe Definitive Guide to Building a Data Mesh with Event StreamsData Mesh 101What is Data Mesh? ft. Zhamak DehghaniData Mesh ArchitectureWatch the video version of this podcastKris Jenkins’ TwitterStreaming Audio Playlist Join the Confluent CommunityLearn more with Kafka tutorials, resources, and guides at Confluent DeveloperUse PODCAST100 to get an additional $100 of free Confluent Cloud usage (details)