The time series preprocessing feature allows you to perform timeline processing and time series analysis during the writing process. You can use Time Series Database (TSDB) to query the results of time series preprocessing. This allows TSDB to provide quick responses without the need to scan raw data points.
The time series preprocessing feature offers an optimal solution for you to query large amounts of time series data. This feature allows you to receive responses in seconds and helps you ensure the stability of the response time.
You can use this feature to preprocess the result sets of time series data queries. This provides a quick method for TSDB to return query results.
- Rollup: A rollup is defined as a single time series that is aggregated over time. A rollup is also known as time-based aggregation. For example, you aggregate the time series data that is collected based on seconds into the data that is organized based on minutes or hours.
- Pre-aggregate: Raw time series data of multiple timelines or metrics is aggregated into the data of a new metric. In most cases, downsampling is implemented on the pre-aggregation results. Downsampling allows you to aggregate data based on coarse-grained time intervals.
The application scenarios include data center monitoring, management of a large number of IoT devices, and computational advertising. In these scenarios, you may need to perform wide time span queries that scan millions of data points or tens of millions of data points. As a result, delays of dozens of seconds are produced during the near real-time data downsampling and aggregation of time series databases. This may cause unstable running or downtime of the database instances. The time series preprocessing feature allows you to create rollups or pre-aggregates, and write the processing results to TSDB. This ensures that TSDB can respond to wide time span queries in a few seconds.
You can use TSDB to perform general time series analysis by using the specified rules. This eliminates the need to capture time series data at the business layer for secondary analysis. This also minimizes the impact of data capturing on TSDB performance, increases the instance throughput for queries, and ensures the stability of the response time.
If the time series preprocessing feature is available for your instance type, you can specify preprocessing rules in the TSDB console to use this feature.