e****7
发帖数: 4387 | 1 OLAP，online analytical processing
1. 谁在用
2. 用那家的
3. 有什么好处
我主要用MS的SSAS，挺容易上手，功能还算全，works well with MS SQL SERVER AND
OTHER MS APP. 性能也不错。
接触过pentaho 的OLAP, 比MS差不少
知道呢Oracle 买了Hyperion , 不知道怎么样了 | E*****e
发帖数: 197 | 2 我主要用MS的SSAS,建系统，很好用。
也用SAP BW, 作为用户，感觉还行 | c*****e
发帖数: 1106 | 3 There are ROLAP and MOLAP, and some offerings try to combine them.
This thread seems to be more about a front-end mdx-based MOLAP.
There are logical MOLAP and physical(storage wise) MOLAP.
A physical MOLAP is well suited to build apps where you have a well defined
and finite space to perform analysis.
The good thing is that measures can be fully computed and stored for the
entire cube space to achieve great performance. Comparing to ROLAP, it
automatically handles the setup, management and navigation of aggregates for
dimension members in the cube. The other good thing is that it enables
writeback above leaf/detail data to do what-if analysis.
The bad thing is that there is increasing difficulty to handle bigger/more
dimensions. It is also not designed to do ad-hoc/dynamic aggregation,
something important when reporting on attributes or custom expressions.
Microsoft has done a good job with analysis service. Their UDM model is a
bit messy, but they certainly put a lot of thoughts and built a lot of
features for a platform.
Hyperion Essbase has a more pure MOLAP model, so it is more profoundly
impacted by the strength and weakness mentioned above. It has a strong
calculation engine, and is behind some of its popular apps.
SAP BW is there to work with other SAP apps, I wouldn't consider it a MOLAP
platform. SAP Hana can take mdx, but I don't know much about supported
models. |
|
