Multidimensional | Tabular | |
Query Performance | If queries are related with pre-aggregated data in the disk, the response performance is equivalent to the Tabular or even better in the case of being cached. | Although aggregations are calculated at query time, the data is read from the memory which makes responses perform better either for aggregated data or more fine-grained data (with the exception mentioned on the left). |
Processing Performance | There are processing dependencies (first we have to process dimensions and then the measure groups); aggregations must be recalculated when dimensions processing occurs. | There are no processing dependencies, since the model is made of “only tables”. Unlike multidimensional model, the processing of partitions can’t be done in parallel. |
Performance Tuning | Performance can be improved by defining aggregations and implementing mechanisms to warmup the cache. | The best way to increase performance is to maximize the available server RAM. |
Data Model | Out-of-the-box features for implementing complex models: many-to-many and parent-child relationships, advanced business logic (through MDX). | Supports models with low/medium complexity, without the need of create relationships between dimensions and facts. Some advanced modeling needs (such as many-to-many relationships) are possible via DAX but it’s harder to implement. |
Development Experience | The development takes longer and it requires more advanced skills (eg MDX). | The development is faster and therefore is often recommended for proof of concepts. The learning curve is also easier because of the similarity between DAX and Excel formulas language, which becomes more accessible to business people. Moreover, you can directly import models built in PowerPivot. |
Scalability | Appropriate for models with a large volume of data. | Conditioned by server’s memory, this approach is more suitableFor small/medium size models. |
Exclusive Features | Exclusive out-of-the-box features: Named-Sets, Actions, Translations, Write-back, Data Mining, Role-Playing Dimensions, Custom Assemblies, Custom Rollups, Display Folders, custom formatation of measures, default values for dimensions attributes. | Although some features (as Actions, Translations and Display Folders) are not out-of-the-box, they are possible by editing the generated XMLA or by using BIDS Helper tool available at CodePlex. |
Security | Advanced security features that can go up to cell-level. It’s possible to set drill through permission at role level. | Row-level security.Drill through permission is set to all roles. |
SQL Server Versions | Available since SQL Server 7.0 (1999)In SQL Server 2012 is available in Enterprise, Business Intelligence and Standard editions. | SQL Server 2012 Enterprise and Business Intelligence editions. |
Client Applications Support | Microsoft Excel, SQL Server Reporting Services, PerformancePoint, PowerView (since Multidimensional now supports DAX queries), etc. | Tabular also supports both MDX and DAX queries. Therefore all client applications that support multidimensional model also supports tabular one. |