Parallel Database Management Systems
Parallel DBMS
A DBMS that runs across multiple processors and disks that have been designed to execute operations in parallel when ever possible in order to improve the DBMS performance.
Parallel DBMS are based on the premise that a single processor system can no longer meet the growing requirements for cost effective scale-ability reliability or performance.
A financially attractive alternative to a single processor driven DBMS would be a parallel DBMS driven by multiple CPU's.
Three main architectures
- Shared Memory
- Shared Disk
- Shared Nothing
Shared Memory
Multiple CPU's share system memory
Shared Nothing
(MPP) Massively Parallel Processing is a multiple processor architecture in which each processor is part of the complete system, with its own memory and disk storage. This architecture is superior to shared memory because its more scaleable and can support multiple CPU's. Performance is optimal when data is requested locally.
No comments:
Post a Comment