Auto retry circuit breaker for enhanced performance in microservice applications

Abstract

Cloud-hosted distributed systems are much more resilient to failure. Resilience is the key factor in avoiding cascading failures in microservice architectures, where circuit-breaker, retry, bulkhead, and rate-limiter patterns are implemented. Through the implementation of these patterns, availability and reliability factors can be greatly improved. Circuit-breaker emphasizes a fail-fast mechanism, so it is highly recommended. Although they are beneficial, some factors, such as configuration, need to be improved to increase the system's throughput. In this paper, an approach called auto retry circuit breaker (ARCB) is implemented, which is a modified version of existing patterns of circuit breaker, and it has been found that the throughput of the system can be increased by 36% when compared to the existing manual circuit breaker pattern.