In today''s OLAP system, integrating fast changing data (e.g. stock data)physically into a cube could be complex and time-consuming, The XML technology today makes it very possible that this data is available in XML databases. Thus, making XML data logically federated into OLAP systems is greatly needed. In previous work, an approach to the logical OLAP-XML federation has been developed, which includes the data models, a SQLXM query language, querying techniques, physical algebra and a prototype of OLAP-XML query engine. In this paper, the OLAP-XML Query Engine now has been integrated with practical query optimization techniques. A rule-based and cost-based query optimizer generates a plan space given the initial plan, and selects the least cost plan in the plan space.New operators and transformation rules are introduced to generate plans. Pruning techniques such as Branch and Bound are used to reduce the plan space. An optimization technique, inlining, is integrated in physical algebra and supported by the new OLAP-XML query engine. Experiments about effectiveness of query optimizations techniques are performed on the new engine, indicating that the optimizations have effectively boosted up the querying performance and greatly improved the logical federation of OLAP and XML systems.