Rewriting Aggregate Queries Using Functional Dependencies within the Cloud

Since many years, companies and laboratories have had a pressing need for processing large amounts of data in areas such as astronomy, medicine or social networks. Cloud computing provides users with a virtually infinite amount of computing resources. Scaling up cloud performance can be usually achieved by using more numerous and/or more powerful nodes. However, this results in high costs as well as using more resources than necessary. In the area of databases, caching and query rewriting are two important ways to improve performance. This paper proposes rewriting rules for aggregate queries using semantic caching in the cloud. We have implemented our proposal in the Pig system and conducted experiments in a private cloud.