多维数据集分区比SSAS多维数据集进程花费更多时间 [英] Cube Partitioning taking more time than usual for SSAS cube process

问题描述

以前使用SSIS包进行SSAS数据库处理（分析服务处理任务）需要6个小时才能完成同步。

<我尝试将该DB中的所有多维数据集分区为3个分区，即1年，2年，然后全部休息。并且还创建了如下的SSIS作业，并将多个分区平行分配到8个SSIS任务中，以便它可以在处理其他
分区时并行处理：

1st。它将为最新的1年数据挑选1年分区，

2nd。它将为最新的2年数据挑选1年和2年分区，

第3。它将选择1年，2年并为所有数据休息所有分区。

我执行了第3个SSIS作业，实际上首先处理所有维度，然后是8个并行任务中的所有分区然后将db同步到查询服务器。但它花了超过12个小时。

在以前的处理中，SSIS任务只用于选择SSAS DB名称，但现在我用来选择任务中的每个SSAS对象以使其运行并行。

由于第三个作业执行所有SSAS DB对象，就像以前处理整个ssas DB一样，我期待它至少应该在同一时间完成（6个小时），但花了两倍。

不知道出了什么问题，任何人都可以请你帮忙。任何信息/建议都会有所帮助

问候，

解决方案

< blockquote>

HI，

您是否尝试多批并行处理多个分区？或单个批次中的多个分区？ - >你对处理任务有一些锁定机制。您是否使用Profiler检查过您的作业是否并行执行？

Hi,

Previously SSAS DB processing using SSIS package(analytics Services processing task) was taking 6 hours to complete with synchronization.

I tried and partitioned all cubes in that DB into 3 partitions i.e. 1 year, 2 year and rest all. And also created SSIS jobs as below and parallely distributed many partitions into 8 SSIS task so that it also can process in parallel while processing other partitions:

1st. it will pick 1 year partitions for latest 1 year data,

2nd. it will pick 1 year and 2 year partitions for latest 2 year data,

3rd. It will pick 1 year, 2 year and rest all partitions for all of the data.

I executed 3rd SSIS job, which actually first processing all dimensions, then all partitions in 8 parallel tasks and then synchronizing db to query server. But it is taking more than 12 hours.

In previous processing, SSIS task used to select only SSAS DB name, but now I used to select every SSAS object within the task to make it run parallel.

Since 3rd job executing all SSAS DB objects, like previously it was processing whole ssas DB, I was expecting atleast it should be completed in same time (6 hours) but it took double.

Not sure what went wrong , can anyone please assist. any info/suggestion will be helpful

Regards,

解决方案

HI,

Are u trying to process multiple partitions in parallel in multiple batches? or multiple partitions in a single batch?-> u have some locking mechanism on processing tasks. Have u checked with Profiler, whether your jobs execute in parallel?

这篇关于多维数据集分区比SSAS多维数据集进程花费更多时间的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！