A time series database system based on columnar storage

Abstract

The invention discloses a time series database system based on columnar storage, comprising a SQL parser: converting SQL into an abstract syntax tree through parsing; a logic plan generator: generating a logical query plan by using the abstract syntax tree, and the logical query plan is composed of relational algebra Representation; query optimizer: optimize the query plan by using the state information of the database, and output the rewritten query plan; executor: execute the query plan in a block iterative manner, responsible for interacting with the storage layer and obtaining the data queried by the user; storage engine : Store data in row+column format. The invention uses the column method to store the data in the database, and realizes the characteristics of high-speed data insertion and the index function of the traditional database based on the combination of column storage and rocksdb. In terms of the executor, some operators are improved to deeply combine with the characteristics of the storage engine. , while taking advantage of the new features of modern processors for query acceleration.

Description

technical field [0001] The invention belongs to the field of data storage and processing, and in particular relates to the design of a time series database system based on columnar storage. Background technique [0002] At present, there are several time series databases on the market, such as influxDB, etc., but most of these databases use LSMtree-based storage solutions. For example, influxDB uses a variant of LSMtree, TSM. The advantage of LSMtree is that the writing speed is fast, but due to the need for hierarchical search , its read data performance is not ideal. SUMMARY OF THE INVENTION [0003] Considering the characteristics of time series data (multiple writes and multiple reads, but few updates and deletions), the present invention proposes a columnar storage-based storage engine and query executor on the traditional database architecture, which can improve the query efficiency of the database system. [0004] The object of the present invention is to be achiev...

AI Technical Summary

Problems solved by technology

[0002] At present, there are several time-series databases on the market, such as influxDB, etc., but most of these databases use LSMtree-based storage solutions. For example, influxDB uses the variant TSM of LSMtree. The advantage of LSMtree is that the writing speed is fast, but due to layered search , its read data performance is not ideal

Images