Design of 5G Architecture Enhancements for Supporting Serverless Computing

초록

Serverless computing is a recent emerged trending technology in the 5G edge computing landscape. Serverless computing's on-demand dynamic scalability allows efficient utilization of limited available resources at the edge. To realize the benefit serverless edge computing, several solutions have been recently proposed to optimize serverless service placement, scaling and scheduling at the edge. However, lack of studies consider the involvement of 5G mobile network in orchestrating connection between requests from user equipments (UE) and serverless edge services. 5G control plane need to be aware of the serverless edge service deployment information determined by those optimized algorithms to avoid serverless service cold start problem caused by steering traffic to unoptimized edge sites. To address this issue, this study introduces enhancements to the current 5G architecture and procedures based on computing-aware traffic steering concept. We proposed a new computing-aware traffic steering controller that can manage serverless edge computing deployment information and influence 5G control plane to configure optimal routing path to serverless applications. The proposed solution allows 5G control plane to efficiently co-operate with advanced serverless deployment solutions to orchestrate serverless service serving at the 5G edge. We evaluated our solution against the normal 5G architecture and the 5G architecture enhancements for edge computing standardized by the 3rd Generation Partnership Project (3GPP). Our solution reduced the number of requests affected by cold-start from 8 to 35% and the edge resource consumption amount from 10 to 41%.

키워드