SAND: Analytic User Scale
Data warehousing applications require a database system capable of efficiently performing ad hoc queries on large amounts of data. Traditional systems designed for OLTP applications with conventional lock-based schedulers do not perform well in this environment. For a query that visits every row in a table, a lock-based scheduler must read-lock all the rows, decreasing concurrency by blocking users who want to write. With a traditional lock-based scheduler, as the number of transactions attempting to execute concurrently increases, the number of transactions able to do any work decreases. Writes may only be done by these systems at load time.
This paper describes SAND’s Generation Based Concurrency Control (GBCC), which is designed to handle the type of queries found in data warehousing. GBCC is an optimistic method of concurrency control which gives users the power to read and write to the database without the overhead incurred by a lock-based system.
Get the white paper
To receive the white paper by return email, please complete and submit the following form: