This document discusses SQL Server 2008 as a platform for managing petabyte-scale data. It defines what a very large database (VLDB) is and notes the challenges in managing massive amounts of data insertion, queries and high availability. The key design philosophy is to partition large databases into smaller, more manageable components rather than having a single large database. Methods for partitioning include by server, instance, database and tables. Partitioning can be by data, such as by month or state, or by function, such as sales vs manufacturing data. Table partitioning commonly uses a "sliding window" approach. Scalability is achieved through clustering while cost is reduced via compression, smaller partitions and moving historical data to cheaper storage.