Concerto: A High Concurrency Key-Value Store with Integrity
Title | Concerto: A High Concurrency Key-Value Store with Integrity |
Publication Type | Conference Paper |
Year of Publication | 2017 |
Authors | Arasu, Arvind, Eguro, Ken, Kaushik, Raghav, Kossmann, Donald, Meng, Pingfan, Pandey, Vineet, Ramamurthy, Ravi |
Conference Name | Proceedings of the 2017 ACM International Conference on Management of Data |
Publisher | ACM |
Conference Location | New York, NY, USA |
ISBN Number | 978-1-4503-4197-4 |
Keywords | composability, Concurrency, indexing, integrity, key-value stores, main memory, merkle trees, Metrics, Outsourced Database Integrity, pubcrawl, recovery, Resiliency, secure hardware, security, verification |
Abstract | Verifying the integrity of outsourced data is a classic, well-studied problem. However current techniques have fundamental performance and concurrency limitations for update-heavy workloads. In this paper, we investigate the potential advantages of deferred and batched verification rather than the per-operation verification used in prior work. We present Concerto, a comprehensive key-value store designed around this idea. Using Concerto, we argue that deferred verification preserves the utility of online verification and improves concurrency resulting in orders-of-magnitude performance improvement. On standard benchmarks, the performance of Concerto is within a factor of two when compared to state-of-the-art key-value stores without integrity. |
URL | http://doi.acm.org/10.1145/3035918.3064030 |
DOI | 10.1145/3035918.3064030 |
Citation Key | arasu_concerto:_2017 |