SMP vs MPP

Symmetric Multi-Processing (SMP)

Symmetric Multi-Processing (SMP) is the primary parallel architecture employed in servers. An SMP architecture is a tightly coupled multiprocessor system, where processors share a single copy of the operating system (OS) and resources that often include a common bus, memory and an I/O system.

Think of this as a typical single server, multi-core with locally attached storage, running the Microsoft Windows OS.

Massively Parallel Processing (MPP)

 Massively Parallel Processing (MPP) is the coordinated processing of a single task by multiple processers, each working on a different part of the task. With each processor using its own operating system (OS) and memory. MPP processors communicate between each other using some form of messaging interface via an “interconnect”. The setup for MPP is more complicated than SMP and one approach to simplify this while providing equal amounts of resources between processors is the “Shared-Nothing” architecture. This approach is the one Parallel Data Warehouse is based upon.

The term shared nothing architecture was coined by Michael Stonebraker (1986) to describe a multiprocessor database management system in which neither memory nor disk storage is shared among the processors.

For a database which follows the shared-nothing architecture, each processor has its own set of disks. Data is “horizontally partitioned” across nodes, such that each node has a subset of the rows from each table in the database. Each node is then responsible for processing only the rows on its own disks. Such architectures are especially well suited to data warehouse workloads, where large fact tables can be distributed across the nodes.

In addition, every node maintains its own lock table and buffer pool, eliminating the need for complicated locking and software or hardware consistency mechanisms. Because shared nothing does not typically have a severe bus or resource contention it can be made to scale to hundreds or even thousands of machines.