Pacer: Comprehensive Network Side-Channel Mitigation in the Cloud

TL;DR

Pacer is a system designed to eliminate network side-channel leaks in public cloud environments by shaping traffic to prevent secret information from being inferred through packet timing and sizes, while maintaining network performance.

Contribution

It introduces a novel traffic shaping technique that prevents internal and external side-channel leaks, with a formal security proof and practical implementation in hypervisor extensions.

Findings

Effective in thwarting CNN-based traffic classification attacks

Imposes moderate overheads on bandwidth, latency, and throughput

Provides end-to-end network side-channel mitigation in public clouds

Abstract

Network side channels (NSCs) leak secrets through packet timing and packet sizes. They are of particular concern in public IaaS Clouds, where any tenant may be able to colocate and indirectly observe a victim's traffic shape. We present Pacer, the first system that eliminates NSC leaks in public IaaS Clouds end-to-end. It builds on the principled technique of shaping guest traffic outside the guest to make the traffic shape independent of secrets by design. However, Pacer also addresses important concerns that have not been considered in prior work -- it prevents internal side-channel leaks from affecting reshaped traffic, and it respects network flow control, congestion control and loss recovery signals. Pacer is implemented as a paravirtualizing extension to the host hypervisor, requiring modest changes to the hypervisor and the guest kernel, and only optional, minimal changes to applications. We present Pacer's key abstraction of a cloaked tunnel, describe its design and implementation, prove the security of important design aspects through a formal model, and show through an experimental evaluation that Pacer imposes moderate overheads on bandwidth, client latency, and server throughput, while thwarting attacks based on state-of-the-art CNN classifiers.