Shared Nothing vs. Shared Everything

Greenplum Database utilizes a shared nothing, massively parallel processing architecture that has been designed for business intelligence. Most of today’s general-purpose relational database management systems are designed for Online Transaction Processing (OLTP) applications. By default, business intelligence applications have inherited this less than optimal architecture. The reality is that BI workloads are fundamentally different from OLTP transaction workloads and therefore require a profoundly different architecture.

OLTP transaction workloads require quick access and updates to a small set of records. This work is typically performed in a localized area on disk, with one or a small number of parallel units. Shared-everything architectures, in which processors share a single large disk and memory, are well suited to OLTP workloads.

However, shared-everything architectures are quickly overwhelmed by the full table scans, multiple complex table joins, sorting, and aggregation operations against vast volumes of data that represent the lion’s share of BI workloads. OLTP architectures perform poorly with BI workloads. Greenplum Database was designed from the ground up as a shared-nothing architecture that is optimal for BI workloads.

Greenplum Database's shared nothing architecture, which leverages a number of self-contained, parallel processing units, is a proven and effective solution to support large-scale data warehousing demands. In shared-nothing architectures like Greenplum Database's, each unit acts as a self-contained database management system that owns and manages a distinct portion of the overall data. Shared-nothing databases automatically distribute data and parallelize query workloads across all available hardware. As a result, shared-nothing databases dramatically out-perform general purpose database systems for BI workloads.