To handle big data volumes we should remember about perfecting data model. Sometimes adding some flags or moveing field from dimension into fact table causes, that QV can better use efficient mechanisms. It may be also worth to consider divide big application into few smaller and use chaining execution.
In bigger solutions it is also very important to have enough RAM for cache.
