Wormhole: A Fast Ordered Index for In-memory Data Management
share
In-memory data management systems, such as key-value store, have become an essential infrastructure in today's big-data processing and cloud computing. They rely on efficient index data structures to access data. While unordered indexes, such as hash tables, can perform point search in the O(1) time, they cannot be used in many scenarios where range operations are required. Many ordered indexes, such as B+ tree and skip list, have a O(log N) lookup cost, where N is the number of keys in the index. For an ordered index hosting billions of keys, it may take more than 30 key-comparisons for one lookup, which is an order of magnitude more expensive than that in a hash table. With large memory and fast network in today's data centers, this O(log N) time is taking a big toll on applications that rely on ordered indexes. In this paper we introduce a new ordered index data structure, named Wormhole, that takes O(log L) worst-case time for looking up a key with a length of L. The low cost is achieved by simultaneously leveraging strengths of three data structures, namely hash table, prefix tree, and B+ tree, to orchestrate a single fast ordered index. Wormhole's range operations can be performed by linear scanning on a list after an initial lookup. This improvement of access efficiency does not come at a price of compromised space efficiency. Instead, Wormhole's index space is comparable to those of B+ tree and skip list. Experiment results show that Wormhole outperforms skip list and B+ tree by up to 8.4x and 4.9x in terms of key lookup throughput, respectively. Its lookup throughput is 1.6x to 6.6x as high as that of Masstree, a highly optimized key-value index.
READ FULL TEXT