基于查询频率的混合事务分析处理数据同步优化

doi:10.3969/j.issn.1000-5641.2022.05.003

摘要/Abstract

摘要：

混合事务分析处理 (Hybrid Transaction Analytical Processing, HTAP) 系统需要同时支持事务处理和查询分析两种工作负载. 为了消除这两种负载之间的干扰, HTAP系统会为这两种负载指定不同的数据副本, 分别处理OLTP (Online Transaction Processing) 和OLAP (Online Analytical Processing) 请求, 并基于日志回放进行副本之间的数据同步. HTAP系统致力于同步OLTP数据到OLAP端以提供更新鲜的数据查询服务, 日志的发送与回放速度是影响数据新鲜度的关键因素. 本文在基于表分组的日志并行回放方法的基础上, 针对HTAP负载特点, 提出基于OLAP端查询频率的日志发送与回放方法. 在保障数据一致性的前提下, 本文所提出的方法提升了高频查询表日志的处理优先级, 最终实现高效的日志发送、日志回放以及针对性的高频查询表数据优先展示, 保证了HTAP系统数据的新鲜度.

关键词: 混合事务分析处理, 数据新鲜度, 日志回放, 冲突检测

Abstract:

A hybrid transaction analytical processing (HTAP) system must concurrently support both transaction processing and query analysis. To eliminate interference between them, HTAP systems also typically assign different copies of data to both workloads, handling online transaction processing (OLTP) and online analytical processing (OLAP) requests separately, and synchronizing data between the copies based on a log replay. An HTAP system is committed to efficiently synchronizing OLTP data to OLAP, thereby providing a fresher data access service. In addition, the speed of sending and replaying the logs of the tables to be queried is a key factor affecting the freshness of the data. In this paper, using the table grouping based log parallel replay method and the characteristics of the HTAP load, a log sending and replay method is proposed based on the query frequency of the OLAP side. To ensure data consistency, this method improves the processing priority of high-frequency query table logs and achieves efficient log sending and replay capabilities along with a targeted priority display of high-frequency query table data, thereby ensuring the freshness of the HTAP system.

Key words: hybrid transaction analytical processing, data freshness, log replay, conflict detection

中图分类号:

TP392

唐永金, 孙家博, 蔡鹏. 基于查询频率的混合事务分析处理数据同步优化[J]. 华东师范大学学报（自然科学版）, 2022, 2022(5): 26-35.

Yongjin TANG, Jiabo SUN, Peng CAI. Optimization of HTAP data synchronization based on query frequency[J]. Journal of East China Normal University(Natural Science), 2022, 2022(5): 26-35.

图/表 6

参考文献 14

1	RAZA A, CHRYSOGELOS P, ANADIOTIS A C, et al. Adaptive HTAP through elastic resource scheduling [C]// Proceedings of the 2020 ACM SIGMOD International Conference on Management of Data. 2020: 2043-2054.
2	PSAROUDAKIS I, WOLF F, MAY N, et al. Scaling up mixed workloads: A battle of data freshness, flexibility, and scheduling [C]// Proceedings of the Technology Conference on Performance Evaluation and Benchmarking. 2014: 97-112.
3	KANTER J M, VEERAMACHANENI K. Deep feature synthesis: Towards automating data science endeavors [C]// Proceedings of the 2015 IEEE International Conference on Data Science and Advanced Analytics. 2015: 15652837.
4	LUO Y, WANG M, ZHOU H, et al. Autocross: Automatic feature crossing for tabular data in real-world applications [C]// Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2019: 1936-1945.
5	LAMH T, THIEBAUT J M, SINN M, et al. One button machine for automating feature engineering in relational databases [EB/OL]. (2017-06-01)[2022-06-12]. https://arxiv.org/pdf/1706.00327.pdf.
6	孙家博. HTAP系统中的并行日志回放优化 [D]. 上海: 华东师范大学, 2022.
7	MAKRESHANSKI D, GICEVA J, BARTHELS C, et al. BatchDB: Efficient isolated execution of hybrid OLTP + OLAP workloads for interactive applications [C]// Proceedings of the 2017 ACM International Conference on Management of Data. 2017: 37-50.
8	LAHIRI T, CHAVAN S, COLGAN M, et al. Oracle database in-memory: A dual format in-memory database [C]// Proceedings of the 2015 IEEE 31st International Conference on Data Engineering. 2015: 1253-1258.
9	LARSON P Å, BIRKA A, HANSON E N, et al. Real-time analytical processing with SQL Server. Proceedings of the VLDB Endowment, 2015, 8 (12): 1740- 1751.
10	YANG J, RAE I, XU J, et al. F1 Lightning: HTAP as a service. Proceedings of the VLDB Endowment, 2020, 13 (12): 3313- 3325.
11	HUANG D X, LIU Q, CUI Q, et al. TiDB: A Raft-based HTAP database. Proceedings of the VLDB Endowment, 2020, 13 (12): 3072- 3084.
12	HONG C T, ZHOU D, YANG M, et al. KuaFu: Closing the parallelism gap in database replication [C]// Proceedings of the IEEE 29th International Conference on Data Engineering. 2013: 1186-1195.
13	XIA Y, YU X Y, PAVLO A, et al. Taurus: Lightweight parallel logging for in-memory database management systems (extended version) [EB/OL]. (2020-10-14)[2022-06-08]. https://arxiv.org/pdf/2010.06760.pdf.
14	QIN D, BROWNA D, GOEL A. Scalable replay-based replication for fast databases. Proceedings of the VLDB Endowment, 2017, 10 (13): 2025- 2036.

[1]	简炜, 胡梓锐, 张蓉. OLAP查询基数预估能力评估[J]. 华东师范大学学报（自然科学版）, 2024, 2024(5): 141-151.
[2]	凌向荣, 翁思扬, 张蓉. 基于锁避免的事务内阻塞分析与调度策略[J]. 华东师范大学学报（自然科学版）, 2024, 2024(5): 152-161.
[3]	连薛超, 刘维, 王清帅, 张蓉. 面向高冲突事务处理的架构设计和优化[J]. 华东师范大学学报（自然科学版）, 2023, 2023(6): 28-38.
[4]	王聪聪, 胡卉芪. 基于持久化内存和共享缓存架构的高性能数据库[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 1-10.
[5]	俞融, 杨攀飞, 王清帅, 张蓉. 数据同步机制自适应优化的HTAP数据库原型系统[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 11-25.
[6]	梅文娟, 蔡鹏. 面向存算分离架构的混合粒度缓存策略[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 26-39.
[7]	刘丹琪, 蔡鹏. 存算分离架构下Part元数据的单独管理策略[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 40-50.
[8]	卢皙钰, 刘维, 翁思扬, 李可强, 张蓉. 基于模糊测试生成多样化的数据库隔离级别测试案例[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 51-64.
[9]	高歌, 胡卉芪. FeaDB: 基于内存的多版本在线特征存储[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 65-76.
[10]	陈现森, 徐辰. 基于openGauss的异构算子加速技术[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 90-99.
[11]	张帅, 胡卉芪, 徐尧强, 周烜. 面向异常检测的时序数据库查询优化[J]. 华东师范大学学报（自然科学版）, 2023, 2023(2): 119-131.
[12]	官嘉林, 朱艳, 吴庭亮, 陈艳, 张敬伟. 基于大页内存的学习索引内存分配策略[J]. 华东师范大学学报（自然科学版）, 2023, 2023(2): 73-81.
[13]	刘梦男, 许建秋. 基于查询代价的两级轨迹数据划分算法[J]. 华东师范大学学报（自然科学版）, 2022, 2022(5): 184-194.
[14]	陈婷, 项兆坤, 徐金凯, 张蓉. 查询优化器连接顺序评估[J]. 华东师范大学学报（自然科学版）, 2022, 2022(5): 48-60.
[15]	游舒泓, 苏仟, 张蓉. 云数据库运行环境的动态模拟[J]. 华东师范大学学报（自然科学版）, 2022, 2022(5): 73-89.