Скачать или смотреть Reliability of ATM processing system

Over the course of two years, our bank transitioned the self-service device fleet (48,000 units of equipment) to our in-house processing, abandoning the vendor SmartVista. During the transition process, we encountered a number of unusual failures in the operation of equipment and open-source products. Analyzing the root causes and options for mitigating the risks of recurring failures, we realized that the solution did not meet all internal requirements. Consequently, we decided to change the system architecture to ensure greater reliability. High-availability PostgreSQL clusters, Stand-In, GreenField, sharding, three levels of caching, and other solutions were implemented to mitigate unavailability risks and reduce potential failure zones.

Evgeny Akimov
Sber