Hybris: Robust hybrid cloud storage

Dobre, Dan; Viotti, Paolo; Vukolic;, Marko
SOCC 2014, 5th annual ACM Symposium on Cloud Computing, November 3-5, Seattle, WA, USA

Besides well-known benefits, commodity cloud storage also raises concerns that include security, reliability, and consistency. We present Hybris key alue store, the first robust hybrid cloud storage system, aiming at addressing these concerns leveraging both private and public cloud resources. Hybris robustly replicates metadata on trusted private premises (private cloud), separately from data which is dispersed (using replication or erasure coding) across multiple untrusted public clouds. Hybris maintains metadata stored
on private premises at the order of few dozens of bytes per key, avoiding the scalability bottleneck at the private cloud. In turn, the hybrid design allows Hybris to efficiently and robustly tolerate cloud outages, but also potential malice in clouds without overhead. Namely, to tolerate up to f malicious clouds, in the common case of the Hybris variant with data replication, writes replicate data across f + 1 clouds, whereas reads involve a single cloud. In the worst case, only up to f additional clouds are used. This is considerably better than earlier multi-cloud storage systems that required costly 3 f + 1 clouds to mask f potentially malicious clouds. Finally, Hybris leverages strong metadata consistency to guarantee to Hybris applications strong data consistency without any modifications to the eventually consistent public clouds. We implemented Hybris in Java and evaluated it using a series of micro and macrobenchmarks. Our results show that Hybris significantly outperforms comparable multi-cloud storage systems and approaches the performance of barebone commodity public cloud storage.

DOI
Type:
Conférence
City:
Seattle
Date:
2014-11-03
Department:
Sécurité numérique
Eurecom Ref:
4414
Copyright:
© ACM, 2014. This is the author's version of the work. It is posted here by permission of ACM for your personal use. Not for redistribution. The definitive version was published in SOCC 2014, 5th annual ACM Symposium on Cloud Computing, November 3-5, Seattle, WA, USA http://dx.doi.org/10.1145/2670979.2670991
See also:

PERMALINK : https://www.eurecom.fr/publication/4414