Se-Lambda: Securing Privacy-Sensitive Serverless Applications Using SGX Enclave

Serverless computing is an emerging trend in the cloud, which represents a new paradigm for deploying applications and services. In the serverless computing framework, cloud users can deploy arbitrary code and process data on the service runtime. However, as neither cloud users nor cloud providers are trustworthy, serverless computing platform suffers from trust issues caused by both sides. In this paper, we propose a new serverless computing framework called Se-Lambda, which protects the API gateway by using SGX enclave and the service runtime by leveraging a two-way sandbox that combines SGX enclave and WebAssembly sandboxed environment. In the proposed service runtime, users' untrusted code is confined by WebAssembly sandboxed environment, while SGX enclave prevents malicious cloud providers from stealing users' privacy-sensitive data. In addition, we implement a privilege monitoring mechanism in SGX enclave to manage the access control of function modules from users. We implement the prototype of Se-Lambda based on the open source project OpenLambda. The experimental results show that the Se-Lambda imposes a low performance penalty, while buying a significantly increased level of security.