v1.0.0
Compatibility
- None
Features
- Supports full join: multi-table full join.
- Supports streaming aggregation based on time buckets: Streaming aggregation is supported to solve the problem of resource consumption for data downsampling in scenarios with a large amount of data.
- Supports multi-level downsampling: Different down-sampling policies can be provided for historical data in different time ranges to reduce storage costs and improve query performance.
- Supports approximate quantile query: Compared with the exact quantile query, the query efficiency is improved by 2 to 5 times while the query precision is guaranteed.
- Supports holtWinters() function: The holtWinters() function applies the Holt-Winters forecasting method to input tables.
- Supports RPC compression transmission: configurable RPC data compression transmission between nodes, reducing bandwidth and improving data transmission efficiency
Performance enhancements
- Compression algorithm: The combination of Snappy, Gorilla, and RLE compression algorithms is used. Different compression algorithms are selected based on data features to ensure a better data compression rate than using one of the compression algorithms alone.
- Memory optimization: Streaming merges out-of-order files to reduce memory usage.
- Memory optimization: Optimize the index search process, Pre-allocate memory based on scenarios, and In-place index update.
- Query optimization: Simplifies the DAG construction process. The ts-sql transfers the schema and DAG relationship diagram to the ts-store and rebuilds the DAG in the ts-store, reducing the network overhead by 95%.
- Query optimization: Pre-tailoring is performed for queries within a time range, improving query performance in scenarios with a large time span.
- Query optimization: File handles are opened on demand, reducing the handle resource consumption of the operating system in scenarios with massive data files.
- Castor: Retry upon failure and connection pool are used to improve reliability, and more detailed error information can be returned.
What's Changed
- fix: fix query used regexpr with escape by @zhenyuxie in #92
- chore: use new GitHub issue template by @shilinlee in #93
- fix: repeatedly adding shard read lock by @jwcesign in #96
- fix: ts-monitor does not check index directory exists by @7StarH in #98
- fix: index out of range for descending query by @smallYellowCat in #100
- fix: check the error for function
indexBuilder.Scan
by @huguowei1996 in #101 - fix: log print protobuf object error by @lyric315 in #103
- fix: always release the session to pool if the session occurred an error by @nicehiro in #108
- fix: deadlock for spdy when rpc session close repeated by @minazukie in #109
- fix: measurement cannot contain special characters by @andrew-sn in #110
- fix: repair the syntax and logic errors in the Docker startup script by @fx408 in #113
- fix: no error is returned when the index creation process fails by @fEyebrow in #116
- feat: support report usage statistics to openGemini by @wangfei1000 in #115
- fix: add report usage statistics url by @shilinlee in #133
- chore: change image files by @xiangyu5632 in #134
New Contributors
- @jwcesign made their first contribution in #96
- @7StarH made their first contribution in #98
- @smallYellowCat made their first contribution in #100
- @huguowei1996 made their first contribution in #101
- @lyric315 made their first contribution in #103
- @nicehiro made their first contribution in #108
- @minazukie made their first contribution in #109
- @andrew-sn made their first contribution in #110
- @fx408 made their first contribution in #113
- @fEyebrow made their first contribution in #116
- @wangfei1000 made their first contribution in #115
Full Changelog: v0.2.0...v1.0.0