A Collision-Mitigation Cuckoo Hashing Scheme for Large-Scale Storage Systems

摘要: With the rapid growth of amount information, cloud computing servers need to process and analyze large amounts high-dimensional unstructured data timely accurately. This usually requires many query operations. Due simplicity ease use, cuckoo hashing schemes have been widely used in real-world cloud-related applications. However, due potential hash collisions, suffers from endless loops high insertion latency, even risks re-construction entire table. In order address these problems, we propose a cost-efficient scheme, called MinCounter. The idea behind MinCounter is alleviate occurrence by selecting unbusy kicking-out routes. selects “cold” (infrequently accessed), rather than random, buckets handle collisions. We further improve concurrency scheme pursue higher performance adapt concurrent has salient features offering efficient services delivering servers, as well enhancing experiences for users. implemented large-scale testbed examined using three traces. Extensive experimental results demonstrate efficacy efficiency