Ground X Reveals Klaytn Storage Optimization and BApp Development Environment Improvements at if(kakao)2020

Ground X shared efforts to improve the performance and cost of the Klaytn platform and enhance Developer Experience (DX) at if(kakao)2020. In particular, technical approaches to resolve the rapid increase in storage capacity on the main network Cypress and BApp development cases utilizing KAS (Klaytn API Service) drew attention.

Storage Capacity Reduction and Memory Optimization In June 2020, the first year after its launch, the total storage capacity of Cypress, Klaytn's main network, exceeded 800GB. Among this, the Storage trie, which accounted for the largest portion at 651GB, had structural inefficiencies in storing all states from the Genesis block. By switching to a method that stores only the latest state data, capacity can be reduced by 75% to approximately 129GB. Additionally, to address the issue where the existing State trie cache occupied more memory than the actual used space, a new caching method was applied, derived mathematically to determine cache miss rates and validated through experiments to improve memory efficiency and speed.

SDK Integration and BApp Development Support To resolve usability differences between caver-js and caver-java, developed in different languages, an SDK common architecture was designed and implemented. This unified terminology and usage across languages, lowering the learning curve for developers and facilitating future expansion of SDKs for new languages. For developers without blockchain experience, a case study of developing a cafe remote ordering system using KAS was introduced, presenting methods for designing BApps suited to data characteristics and ways to mitigate development difficulty through KAS.

DevOps Infrastructure Automation Ground X established a DevOps culture of managing cloud infrastructure as code by utilizing Kubernetes, Terraform, and Helm. This maintained high development productivity and introduced an AWS Connect-based operations system, building a flexible operational framework capable of responding to incidents without physical computers.