Cuckoo Trie: Exploiting Memory-Level Parallelism for Efficient DRAM Indexing

TL;DR

The paper introduces the Cuckoo Trie, a memory-efficient ordered index structure that leverages memory-level parallelism to significantly improve search performance on modern hardware.

Contribution

It presents the Cuckoo Trie, a novel index structure that exploits memory-level parallelism without sacrificing memory efficiency, outperforming existing indexes.

Findings

Outperforms state-of-the-art indexes by up to 360%.

Achieves faster search times on various datasets.

Maintains comparable or smaller memory footprint.

Abstract

We present the Cuckoo Trie, a fast, memory-efficient ordered index structure. The Cuckoo Trie is designed to have memory-level parallelism -- which a modern out-of-order processor can exploit to execute DRAM accesses in parallel -- without sacrificing memory efficiency. The Cuckoo Trie thus breaks a fundamental performance barrier faced by current indexes, whose bottleneck is a series of dependent pointer-chasing DRAM accesses -- e.g., traversing a search tree path -- which the processor cannot parallelize. Our evaluation shows that the Cuckoo Trie outperforms state-of-the-art-indexes by up to 20%--360% on a variety of datasets and workloads, typically with a smaller or comparable memory footprint.