Use multiple marts to create, maintain, and use more than one schema for your Data Marts. Different schemas share a single Eagle data warehouse as their source of data. You control all of your multiple Data Marts from the Data Mart component of PACE.

Should You use Multiple Marts?

The multiple marts option offers you many possibilities for information delivery. They can help you scale your data delivery as your business grows.

However, multiple marts add complexity along with flexibility, and you should use them only after careful analysis and planning. Each mart is configured and updated independently. There is no support in the product for sharing physical tables across marts, so there may be duplication of data that places additional demands upon system resources in the build process. When duplicate data is built, some mart content can temporarily get out of synchronization. Given that each mart is configured separately, it is possible to select a different field attribute for a field in one mart than was selected for that field in another mart. Using the Copy Model feature can greatly mitigate this risk. In addition, firmwide reporting is usually done from just one mart, so some level of data duplication may be unavoidable.

Advantages of Multiple Marts

You use multiple marts for a variety of reasons, including:

• Departmental Marts. Within an organization there may be different teams with very different information and reporting requirements. These teams and the report developers who support them may want to use marts dedicated to each team. Sharing of fields may introduce confusion and conflicting objectives. When this happens, multiple marts can help. However, if most Mart content is actually shared across teams, with relatively small differences among them, dedicated views against the tables of a single mart may be an adequate approach

• Operations in Different Time Zones. A firm with global scope may have money management operations and information delivery workflows in more than one time zone. This may warrant multiple marts if one location needs to update its mart at the same time that another location must report against the mart.

• Load-Splitting for High-volume Mart Builds. In a given Mart schema, each build must generate the overall dmart_fund_id instance (combination of entity, effective date, and snapshot) on a serial basis for all entities, dates, and snapshots involved. This can be a bottleneck factor for a very high-volume build. However, in a high-volume application it is possible to distribute the build across several schemas, reducing overall build time. If this is the requirement, Copy Model can be especially useful to standardize mart structure and content across schemas.

These are some of the considerations involved in the decision to use multiples. Eagle Global Service and Support can help each client evaluate this decision and achieve an overall design that takes best advantage of this powerful capability.

Additional Mart Types

Each mart instance is a distinct destination to which the Data Mart engines can push data. You can maintain as many marts of any of these types as you want. Possible types of destination schema include:

• Marts inside the Eagle data warehouse.

• Marts in databases outside the Eagle data warehouse, using the same database management system as the Eagle warehouse.

• Marts in databases outside the Eagle data warehouse, not using the same database management system as the Eagle warehouse, but either SQL or Oracle.

Note that special provision must be made for marts built in a SQL Server database, where Eagle PACE runs on a UNIX or Linux platform. Refer to Setup for Remote Marts Running from Solaris or Linux to SQL Server for more information.