Taos Data

Hongze Cheng

Jianhui Tao

Haojun Liao

Shengliang Guan

A method for storing time series data is provided, the method includes: caching time series data containing at least one record to a memory in a row-based storage manner; when the time series data needs to be written to a disk, determining a sum of number of records according to the to-be-written-to-disk time series data and time series data in a LAST file of the disk; if the sum is less than a pre-set number N, writing the to-be-written-to-disk time series data to the LAST file of the disk; if the sum is greater than or equal to the pre-set number N, merging the to-be-written-to-disk time series data and the time series data in the LAST file of the disk, and writing merged time series data to a DATA file for permanently saving data of the disk in a column-based storage manner.

Method for storing time series data

The present invention discloses an efficient storage method for time-series data, comprising: storing time-series data collected by each data sampler and static attribute data of its data sampler separately; when each time series is stored, based on a unique ID of each time series, storing corresponding time-series data; and when the static attribute data is stored, the static attribute data are stored in another database independently, wherein each data sampler saves one corresponding record in a database storing the static attribute data. According to the present invention, the storage space can be saved, the processing efficiency and the processing speed can be improved, and the query processing can be flexibly carried out.

Efficient storage method for time series data

The invention discloses an aggregation optimized processing method for time-series data, characterized by comprising the following steps:

Method for aggregation optimization of time series data

The invention discloses a processing method for process changing time-series database table structure, wherein comprising following steps: an application side generates a data table containing a table name ID, a schema version, and a column ID; when a data column is increased, the application side modifies a schema of the data table, increases the schema version, assigns a new column ID number to a newly increased column incrementally, and assigns a default value to the newly increased column; according to a data insertion request of the application side, a data node receives data of a schema version carrying a data table from the application side and writes the received data, wherein writing the received data comprises: storing the schema version of the data table carried by data from the application side by the data node, and writing the received data, and writing the received data; when the data node receives new data from the application side, comparing the schema version of the data table carried by the new data with a stored schema version of the same data table; and the data node writes data according to the comparison result. According to the method, the operation of changing the structure of a table can be completed instantly, the historical data does not need to be changed, new and old table structure definitions can be used in parallel, and the flexibility can be improved.

Growing community of inventors

Beijing, China

Hongze Cheng