Scalable Resource Provisioning For Multi-tenant SaaS With Cloud Functions

Abstract

SaaS cloud services support the sharing of software resources among multiple tenants, but the utilization of computing resources by different tenants is different. Some tenants with large business volumes and large amounts of data tend to occupy more CPU and database resources, especially when the SaaS platform provides computing-intensive services such as machine learning, which will cause more serious resource overhead and seriously affect tenants' experience. Therefore, this paper designs and implements a scalable resource provisioning scheme based on cloud functions for multi-tenant architecture. We present a self-defined scheduling protocol, scheduling center, and distributed nodes to alleviate the central server pressure, rationally scheduling and allocating tenant requests by running functional tenant services across nodes.