In this paper, we present an implementation of a cuckoo filter for membership testing, optimized for distributed data stores operating in high workloads. In large databases, querying becomes inefficient using traditional search methods. To achieve optimal performance it is necessary to use probabilistic data structures to test the membership of a given key, at the cost of getting false positives while querying data. The widely used bloom filters can be used for this, but they have limitations like no support for deletes. To improve upon this we use a modified version of the cuckoo filter that gives better amortized times for search, with less false positives.

中文翻译：

---

优化 Cuckoo 过滤器以获得高突发耐受性、低延迟和高吞吐量

在本文中，我们介绍了一种用于成员资格测试的布谷鸟过滤器的实现，该过滤器针对在高工作负载下运行的分布式数据存储进行了优化。在大型数据库中，使用传统搜索方法进行查询变得效率低下。为了获得最佳性能，必须使用概率数据结构来测试给定键的成员资格，但代价是在查询数据时会出现误报。广泛使用的布隆过滤器可用于此目的，但它们有一些限制，例如不支持删除。为了改进这一点，我们使用了修改版的布谷鸟过滤器，它为搜索提供了更好的分摊时间，减少了误报。