Shared resource entanglement attacks against serverless computing
Published in 2024 IEEE Conference on Communications and Network Security (CNS), 2024
Serverless computing has emerged as a popular paradigm for building and deploying applications in the cloud. At the heart of this paradigm lies the concept of serverless functions, which have gained significant popularity due to their simplicity, scalability, and cost-efficiency. However, running serverless services on the same physical machine can introduce security threats and risks, primarily when identical serverless functions belong to different tenants. In this paper, we investigate the resource consumption of serverless functions and propose a shared resource entanglement (SRE) threat model. The SRE threat model enables adversaries to infer sensitive data from a victim’s serverless function by intercepting and monitoring the host kernel interacting data patterns and the data load and write rate patterns. The experiment results demonstrate that adversaries can identify patterns in the data processing of coresident serverless functions using the SRE threat model. Also, this vulnerability could lead to the leakage of sensitive data in the Kubernetes platform.