Shared-Nothing Architecture

GaussDB(DWS) is an MPP system employing the shared-nothing architecture. It consists of multiple independent logical nodes that do not share system resources with each other, such as CPUs, memory, and storage. In such an architecture, service data is stored on multiple physical nodes. Data analysis tasks are executed in parallel on the nodes where data is stored. The massively parallel data processing significantly improves response speed.

Figure 4-1 compares the shared-nothing architecture and other architecture.

Figure 1 Architecture comparison

The shared-nothing architecture has the following advantages:

• High scalability
  • Provides on-demand scaling for BI and data analysis to process high-concurrency massive data.
  • Provides an automatic parallel processing mechanism.
• Implements automatic internal parallel processing without manual partitioning or optimization.
  • Allows you to load and access data as you would in a common database.
  • Distributes data on all parallel nodes.
  • Every node processes only partial data.
• Optimal I/O processing
  • All nodes process data in parallel.
  • Nodes share nothing with each other and have no I/O conflicts.
• Increased storage space and improved query and loading performance with added nodes