En jämförelse av cache-hantering och prestanda i REST och GraphQL

Sammanfattning

Today the importance of being able to share data quickly and efficiently is invaluable. To meet this need information must be easily accessible via APIs that are easy to integrate with. Data transfer performance is an important part of the architectural choice. A central method for improving performance is caching, which means that previously retrieved data is temporarily stored to reduce the number of repeated server calls. This study examines and measures how caching affects REST and GraphQL using a quantitative experimental design. The aim is to understand how caching differs in REST and GraphQL and how cold and warm cache affects both APIs. The methodology used is quantitative experimental design. The experiments measure response time and transferred data volume in both cold and warm cache scenarios. Two test cases are used with different levels of data complexity. Both APIs are implemented in a local environment using the same dataset, but with different caching strategies and architectural setups. The results show that response time differs between REST and GraphQL depending on data volume and query complexity. With larger datasets and cold cache REST had shorter response times than GraphQL. With warm cache GraphQL showed the greatest improvements in response time, REST also improved but to a lesser extent. The transferred data volume remained constant between cold and warm cache meaning that caching primarily affected response time rather than data volume. GraphQL consistently transferred more data than REST in both test cases, which stands out in the results. This may be explained by how nested queries and relationships in the implementation return full objects and additional structural information in the response. Overall REST showed lower response times with cold cache, whereas GraphQL achieved the largest performance improvements with warm cache.