CRM V3.0 comes with a brand new web service that provides information about the CRM metadata (e.g. what business entities are in the system, their attribute list, picklist values, etc). When we designed this web service, we had a specific goal to make it useable by a metadata cache program so partners can improve the performance of their metadata-intensive apps by building such a cache. Caching the metadata information can make your apps run faster, considering that the metadata is not changed as often as the actual data (e.g. it is changed when the system is customized). The good news is that as part of the RSS application for CRM, Michaeljon has built a metadata cache using the metadata web service. The source code for this cache program is available in the recently released CRM RSS application package from CRM Sandbox. It certainly is a good starting point if you are planning to build your server side metadata cache and boost your CRM add-on performance.