Ensuring Eventual Consistency in a Microservices Architecture

Microservices Community
149 abonnenter
Speaker: Kaj Steen Bromose, UDL, Denmark

Abstract: How can the architecture of transaction oriented business systems be transformed to a microservices architecture in a “transaction safe” way, where it can be ensured that the system ends up in consistent state? The foundation of our work is the CAP theorem. The CAP theorem states that it is impossible for distributed systems to simultaneously provide more than two out of the following three guarantees: Consistency, Availability, Partition tolerance. With respect to the CAP theorem, microservice architectures are in the AP category, leaving the system to deal with eventual consistency. This is a major issue to handle in some business systems, where database consistency is an absolute must. One solutions is the use of the SAGA pattern, where compensating transactions handles inconsistency and eventually ensures consistency. However, there is a serious issue with compensating transactions – they require transactions to be reversible. Yet, many business transactions are irreversible by nature – for example an automated teller machine transaction, where reclaiming the out handed money cannot be guaranteed to be honored. This webinar will introduce a way to ensure eventual consistency in a microservices architecture with irreversible transactions. The webinar will promote using a dynamic CAP approach, where the CAP positioning of the system is dynamic based on the concrete business event. The will webinar present a microservice architectural pattern where the transactional business events are encapsulated in modular micro monoliths. The dynamic CAP approach and the use of modular micro monoliths has been formulated and developed as a joint effort between a Bankdata and UCL University College. Bankdata has a long history of handling eventual consistency with a proven record of success. The dynamic CAP approach is based on these concrete experiences and a more theoretical approach is taken from a CAP perspective.

Date: February 9th, 2022