Characteristics of Local Cubes

Local cubes have dimensions, hierarchies, levels, members, measures, calculated members, and member properties. Local cube have most, but not all of the capabilities of Analysis Services cubes. A user browsing a cube in an OLAP client application normally can't tell if the application is attached to an Analysis Services cube or a local cube.

The defining characteristic of a local cube is its portability. It's not always convenient, practical, or even possible to be connected to the Analysis Server on your network. If you put your multidimensional data in local cubes, you don't have to be connected to the server. You can browse a local cube on your laptop. You can e-mail a local cube to a colleague. You can create a local cube with a particular subset of an Analysis Services cube's data, to tailor the cube precisely for your desired use.

There are situations, of course, where it's better to connect directly to an Analysis Services cube than to a local cube. Local cubes lack three basic features that are available in Analysis Services cubes:

1. Local cubes do not have any provision for security in SQL Server 2000. This is an important limitation, but local cubes can still be a central component in a secure distribution system for multidimensional data. You can create different local cubes for different users or groups of users. You can then control access to these cubes by limiting access to the cube files. SQL Server 2005 and 2008 do provide security for local cubes.

2. Local cubes do not store any aggregations of the data (whether it be SQL Server 2000, 2005, or 2008). The lack of aggregations is noticeable when doing multidimensional browsing with large local cube files. But when smaller or moderate-sized local cubes are used, the browsing speed is excellent. When you create local cubes you can exclude the measures, dimensions, levels, members, and member properties that you don't need. By excluding the portions of the cube you don't need, you can keep your local cubes to a reasonable size and maintain your performance.

3. You cannot do an incremental update on a local cube. The only way to get new data into a local cube is to delete the old file and create a new one. This limitation isn't very serious because local cube files can be created very quickly, especially when the local cubes are created from Analysis Services (OLAP Services) source cubes.

Local cubes can use either MOLAP (Multidimensional OLAP) or ROLAP (Relational OLAP) storage. ROLAP local cubes are smaller and can be created more quickly, but they don't give you the independence provided by the local cube technology - you have to be connected to the data source to use them. MOLAP cubes include all the cube structure and the cube data in one cube file that can be used without having any access to the original source data.

The default extension for a local cube file is .cub.