2019/11/05 by Mazen Alwadi, Alwadi, Mazen, Aziz Mohaisen +3
Computer Science · Engineering · #Advanced Data Storage Technologies #Advanced Memory and Neural Computing #Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Parallel Computing and Optimization Techniques
paper · pdf · doi:10.48550/arxiv.1911.01922
openalex publication_date 2019/11/05 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Emerging Non-Volatile Memories (NVMs) bring a unique challenge to the\nsecurity community, namely persistent security. As NVM-based memories are\nexpected to restore their data after recovery, the security metadata must be\nrecovered as well. However, persisting all affected security metadata on each\nmemory write would significantly degrade performance and exacerbate the write\nendurance problem. Moreover, recovery time can increase significantly (up to\nhours for practical memory sizes) when security metadata are not updated\nstrictly. Counter trees are used in state-of-the-art commercial secure\nprocessors, e.g., Intel's Safe Guard Extension (SGX). Counter trees have a\nunique challenge due to the inability to recover the whole tree from leaves.\nThus, to ensure recoverability, all updates to the tree must be persisted,\nwhich can be tens of additional writes on each write. The state-of-art scheme,\nAnubis, enables recoverability but incurs an additional write per cache\neviction, i.e., reduces lifetime to approximately half. Additionally, Anubis\ndegrades performance significantly in many cases. In this paper, we propose\nPhoenix, a practical novel scheme which relies on elegantly reproducing the\ncache content before a crash, however with minimal overheads. Our evaluation\nresults show that Phoenix reduces persisting security metadata overhead writes\nfrom 87 % extra writes (for Anubis) to less than write-back compared to an\nencrypted system without recovery, thus improving the NVM lifetime by 2x.\nOverall Phoenix performance is better than the baseline, unlike Anubis which\nadds 7.9 % (max of 35 %) performance overhead.\n