有关Analysis Services体系结构的问题(来自SAP世界) [英] Question about Analysis Services Architecture (coming from SAP world)

问题描述

我来自SAP世界，需要了解如何在Azure Analysis Services中构建类似的体系结构，例如SAP BW.
在SAP BW中，您可以将数据直接从源系统中拉到多维数据集中.好吧，实际上您不需要，而是在最终将其加载到多维数据集之前构建一些用于转换等的分层体系结构.但是，它们全部基于相同的RDBMS.

I am coming from the SAP-world and need to understand how I can build a similar architecture in Azure Analysis Services like with SAP BW.
In SAP BW you can pull the data directly from the source systems into a cube. Well, practically you don't, and instead build some layered architecture for transformations etc. before finally loading it into the cube. However, it is all based on the same RDBMS.

现在来看Azure分析服务，我知道关系数据存储(用于转换，例如使用Azure SQL DW)和OLAP(分析服务)是分开的.我使用任何RDBMS来准备数据，然后将其从 Analysis Services(据我所知也是SQL Server的内幕).

Now looking at Azure Analysis Services, I understand that relational data storage (for transformations, e.g using Azure SQL DW) and OLAP (Analysis Services) is separated. I use any RDBMS to prepare the data, and then move it on into the OLAP engine from Analysis Services (which is also an SQL Server under the hood as far as I understand).

那么，如果我已经清除和转换了数据，是否可以直接将ETL传递给Analysis Services，并将数据直接保留在多维数据集中?也许还可以通过某种磁盘卸载机制来省钱.

So, if I already have cleansed and transformed data, can I directly ETL to Analysis Services and keep the data directly in the cube? Maybe also with some kind of disk offloading mechansim for saving money.

那HANA呢?是否有类似于ADSO的内容? (与多维内存相比，扁平柱状内存)

And what about HANA? Is there something similar to an ADSO? (flat columnar in-memory, in contrast to a multi-dimensional in-memory)

推荐答案

在SAP BW中，您可以拉将数据直接从源系统放入多维数据集.好吧，实际上您不需要，而是构建一些分层的 最终将其加载到多维数据集之前进行转换等的体系结构.但是，它们全部基于相同的RDBMS.

In SAP BW you can pull the data directly from the source systems into a cube. Well, practically you don't, and instead build some layered architecture for transformations etc. before finally loading it into the cube. However, it is all based on the same RDBMS.

----在微软世界中，您也可以直接从源系统中提取数据并在其上创建一个多维数据集，但是根据最佳实践和久经考验的体系结构，始终建议并建议先清理和升级数据.转换数据，然后构建 尺寸和事实，以在其上创建一个多维数据集.相信我，这些概念独立于任何技术或工具.

----In microsoft world also, you can directly pull data from source systems and create a cube on top of it, but as per the best practices and proven architecture it is always recommended and suggested to first clean & transform the data and then build dimensions and facts to create a cube on top of it. Trust me these concepts are independent of any technologies or tools. 

因此，如果我已经清理和转换了数据，可以直接将ETL传递给Analysis Services并将数据直接保存在立方体?也许还可以使用某种磁盘卸载机制 省钱.

So, if I already have cleansed and transformed data, can I directly ETL to Analysis Services and keep the data directly in the cube? Maybe also with some kind of disk offloading mechanism for saving money.

那HANA呢?是否有类似于ADSO的内容? (与多维内存相比，扁平柱状内存)

And what about HANA? Is there something similar to an ADSO? (flat columnar in-memory, in contrast to a multi-dimensional in-memory)

----是的，您可以，但是正如我之前所说，在构建多维数据集之前，应将数据建模为维度和事实来自源系统，因为它们是高度规范化的结构，但是如果您不这样做， 想要做到这一点，那么您可以使用SSAS表格模型，因为它们具有记忆功能，并且可以直接连接到关系源系统以构建模型/多维数据集.

---- Yes you can but as i said earlier you should model your data into dimension and facts before building cubes from source systems as those are highly normalized structures, but still if you don't want to do that then you can go with SSAS tabular models as they have columnar in memory capabilities and you can directly connect to relational source systems to build you models/cubes.

这篇关于有关Analysis Services体系结构的问题(来自SAP世界)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！