Self-Tuning, GPU-Accelerated Kernel Density Models for Multidimensional Selectivity Estimation

posted @ 2021-01-21 11:36 fxjwind 阅读(18) 评论(0) 推荐(0) 编辑

论文解析 -- Selectivity Estimation for Range Predicates using Lightweight Models

摘要:摘要 (点题)Query optimizers depend on selectivity estimates of query predicates to produce a good execution plan. (当前的方案)When a query contains multiple pr 阅读全文
posted @ 2021-01-19 15:45 fxjwind 阅读(38) 评论(0) 推荐(0) 编辑

论文解析 -- LeanStore: In-Memory Data Management Beyond Main Memory

摘要:INTRODUCTION Managing large data sets has always been the raison d’ˆetre (a French expression commonly used in English, meaning "reason for being" or 阅读全文
posted @ 2021-01-05 14:55 fxjwind 阅读(68) 评论(0) 推荐(0) 编辑

论文解析 -- Anti-Caching: A New Approach to Database Management System Architecture

摘要:INTRODUCTION (Disk-oriented数据库的问题,维护buffer pool)DBMSs invariably(always) maintain a buffer pool of blocks in main memory for faster access. When an ex 阅读全文
posted @ 2020-12-25 16:08 fxjwind 阅读(43) 评论(0) 推荐(0) 编辑

论文解析 -- What’s Really New with NewSQL?

摘要:ABSTRACT (点题)A new class of database management systems (DBMSs) called NewSQL tout(吹捧) their ability to scale modern on-line transaction processing (O 阅读全文
posted @ 2020-12-23 18:06 fxjwind 阅读(31) 评论(0) 推荐(0) 编辑

论文解析 -- An Empirical Evaluation of In-Memory Multi-Version Concurrency Control

摘要:ABSTRACT (点题)Multi-version concurrency control (MVCC) is currently the most popular transaction management scheme in modern database management system 阅读全文
posted @ 2020-12-16 17:38 fxjwind 阅读(68) 评论(0) 推荐(0) 编辑

论文解析 -- TiDB: A Raftbased HTAP Database

摘要:摘要 Hybrid Transactional and Analytical Processing (HTAP) databases require processing transactional and analytical queries in isolation to remove the 阅读全文
posted @ 2020-12-01 15:50 fxjwind 阅读(145) 评论(0) 推荐(0) 编辑


摘要: 为什么要用列存这里就不聊了,直接看格式的演变 NSM (N-ary Storage Model) ,按行存储 D 阅读全文
posted @ 2020-05-14 14:46 fxjwind 阅读(296) 评论(0) 推荐(0) 编辑

Optimizing Druid with Roaring bitmaps

摘要:这篇主要是说,如何利用compressed bitmap来提升查询性能 虽然之前有很多bitmap的压缩方案,但是,新提出的Roaring bitmap会更加高效 参考,Better bitmap performance with Roaring bitmaps BitMap,为了有效和compac 阅读全文
posted @ 2020-04-27 15:52 fxjwind 阅读(273) 评论(0) 推荐(0) 编辑

Processing a Trillion Cells per Mouse Click

摘要:Google的论文, Google已经有一些大数据系统,都是基于Full Scan 这里PowerDrill,核心利用了skipping技术,可以提升10到100倍的查询性能 这篇论文的题目让人有点摸不着头脑,这里给出了解释, 整体的思路, 就是先skip,然后再full scan 那么就是,他这里 阅读全文
posted @ 2020-04-26 12:16 fxjwind 阅读(208) 评论(0) 推荐(0) 编辑