笔记记下来，划重点:
https://docs.microsoft.com/zh-cn/sql/analysis-services/comparing-tabular-and-multidimensional-solutions-ssas?view=sql-server-2017
Model Features
| Multidimensional | Tabular | |
| Actions | Yes | No |
| Aggregations | Yes | No |
| Calculated Column | No | Yes |
| Calculated Measures | Yes | Yes |
| Calculated Tables | No | Yes1 |
| Custom Assemblies | Yes | No |
| Custom Rollups | Yes | No |
| Default Member | Yes | No |
| Display folders | Yes | Yes1 |
| Distinct Count | Yes | Yes (via DAX) |
| Drillthrough | Yes | Yes (depends on client application) |
| Hierarchies | Yes | Yes |
| KPIs | Yes | Yes |
| Linked objects | Yes | Yes (linked tables) |
| M expressions | No | Yes1 |
| Many-to-many relationships | Yes | No (but there is bi-directional cross filters at 1200 and higher compatibility levels) |
| Named sets | Yes | No |
| Ragged Hierarchies | Yes | Yes1 |
| Parent-child Hierarchies | Yes | Yes (via DAX) |
| Partitions | Yes | Yes |
| Perspectives | Yes | Yes |
| Row-level Security | Yes | Yes |
| Object-level Security | Yes | Yes1 |
| Semi-additive Measures | Yes | Yes |
| Translations | Yes | Yes |
| User-defined Hierarchies | Yes | Yes |
| Writeback | Yes | No |
Data Considerations
Compression
...multidimensional database will be about one third size of the original data. Tabular databases can sometimes get greater amounts of compression, about one tenth the size, especially if most of the data is imported from fact tables.
Size of the model and resource bias (in-memory or disk)
Historically, the largest databases in production are multidimensional, with processing and query workloads running independently on dedicated hardware, each one optimized for its respective use. Tabular databases are catching up quickly, and new advancements in DirectQuery will help close the gap even further.
Security Features
-
Tabular model databases can use row-level security, using role-based permissions.
-
Multidimensional model databases can use dimension and cell-level security, using role-based permissions.
多说一句，总的来说：
多维（multidimensional) ：强大，功能多一些，难用，开发周期长，需要开发人员的技能高些（建模、MDX），靠近程序猿；
表格(tabular)：易用，适合快速开发，靠近Excel 用户，适合微软现在的发展思路，line up with PowerBI等产品线
Tabular