Commercial OLAP systems usually consider OLAP dimensions as static entities. In practice, dimension updates are often necessary in order to adapt the multidimensional database to changing requirements. Query languages are then needed to retrieve historical information from a data warehouse. Although many proposals addressing temporal OLAP exist, no one supports full operation over the World Wide Web. After introducing a temporal data model supporting historical dimensions and fact table versioning, we present a three-tier architecture based on Web services, SOAP and XML, allowing efficient querying over the Web. In this architecture, XML metadata is stored at the application server in the form of XML documents containing the data warehouse structure. This allows addressing most of the requests (navigation through time, updates, and queries) without accessing the data warehouse. We present our implementation, and also discuss query processing issues, like answering historical queries using materialized views.