diff --git a/v2.4.0-release-note.txt b/v2.4.0-release-note.txt index b3df61df..f0dd7ceb 100644 --- a/v2.4.0-release-note.txt +++ b/v2.4.0-release-note.txt @@ -1,21 +1,20 @@ -V2.4.0-release 版本主要修改集中在: -1、分布式计算性能提升: -● hasAggs/having/sort/limit/Stable function/ 优化下推到DN上执行,性能提升近百倍。 -● 分布式进程ProcLock分段锁优化,减少锁冲突,执行效率提升约5倍。 -● 执行计划优化:重分布的节点数为1时跳过remote算子,精简执行计划。 +The main revisions of V2.4.0-release focus on: +1. Distributed computing performance improvement: +● The hasAggs/having/sort/limit/Stable function/ optimization is pushed down to the DN for execution, and the performance is improved nearly a hundred times. +● Distributed process ProcLock segmentation lock optimization reduces lock conflicts and improves execution efficiency by about 5 times. +● Execution plan optimization: When the number of redistributed nodes is 1, the remote operator is skipped and the execution plan is simplified. -2、2PC事务优化: -● 死锁检查性能优化:批量获取gxid;遍历查找gxid修改为二分查找;增加try轮次限制。 -● 2PC添加保护模式,杜绝部分提交的情况。 -● drop database 增加prepare过程,确保不会出现节点失败残留。 +2. 2PC transaction optimization: +● Deadlock checking performance optimization: obtain gxids in batches; change gxids from traversal search to binary search; increase try round limit. +● 2PC adds a protection mode to prevent partial submissions. +● The prepare process is added to drop database to ensure that there will be no residual node failures. -3、易用性增强: -● 增加pooler多线程日志功能,方便用户进行问题分析。 -● GTM日志优化,日志汇聚一个文件,并自动拆分活跃日志,方便用户随时查看。 -● 支持gtm_ctl -l指定日志文件路径启动,方便用户对日志进行管理。 - -4、新特性支持: -● 自研分区表剪枝,添加IN语法剪枝,提前加速剪枝。 -● 数据shuffle支持用户自定义函数。 -● 允许删除带有分布式外键约束的主表的列。 +3. Enhanced usability: +● Added pooler multi-thread log function, which is convenient for users to analyze problems. +● GTM log optimization, logs are aggregated into one file, and active logs are automatically split, which is convenient for users to view at any time. +● Support gtm_ctl -l to specify the log file path to start, which is convenient for users to manage logs. +4. New feature support: +● Self-developed partition table pruning, adding IN syntax pruning, to accelerate pruning in advance. +● Data shuffle supports user-defined functions. +● Allow dropping columns of primary tables with distributed foreign key constraints.