SimEnc - High-Performance Similarity-Preserving Encryption for Encrypted Docker Image Deduplication

Explore a groundbreaking approach to deduplicating encrypted Docker images in this 18-minute conference talk from USENIX ATC '24. Delve into SimEnc, a high-performance similarity-preserving encryption method that addresses the challenges of deduplicating encrypted images in Docker registries. Learn how this innovative technique integrates semantic hash into message-locked encryption (MLE) to extract semantic information among layers, improving deduplication ratios. Discover the fast similarity space selection mechanism and the novel approach of using Huffman decoding instead of complete layer decompression. Gain insights into SimEnc's superior performance compared to both state-of-the-art encrypted serverless platforms and plaintext Docker registries, including its ability to reduce storage consumption by up to 261.7% and 54.2% respectively, while also improving pull latency.

USENIX ATC '24 - SimEnc: A High-Performance Similarity-Preserving Encryption Approach for...