skip book previous and next navigation links
go up to top of book: HP OpenVMS Alpha Partitioning and Galaxy GuideHP OpenVMS Alpha Partitioning and Galaxy Guide
go to previous page: Developing OpenVMS Galaxy ProgramsDeveloping OpenVMS Galaxy Programs
go to next page: OpenVMS NUMA AwarenessOpenVMS NUMA Awareness
end of book navigation links

3NUMA Implications on OpenVMS Applications  



NUMA (nonuniform memory access) is an attribute of a systemin which access time to any given physical memory location is notthe same for all CPUs. Given this architecture, you must have consistentlygood location (but not necessarily 100 percent of the time) forhigh performance.

The operating system treats the hardware as a set of resourceaffinity domains (RADs). A RAD is the software grouping of physicalresources (CPUs, memory, and I/O) with common access characteristics.On the AlphaServer GS80/160/320 systems, a RAD corresponds to aquad building block (QBB); on AlphaServer ES47/ES80/GS1280 systems,a RAD corresponds to a two-processor CPU board. CPUs access memoryin their own RAD faster than they access memory in another RAD.

If OpenVMS is running on the resources of a single RAD, thenthere is no NUMA effect and this discussion does not apply. Wheneverpossible and practical, you can benefit by running in a single RAD,which eliminates the complexities NUMA may present.

The most common question for overall system performance ina NUMA environment is, "uniform for all?" or "optimalfor a few?" In other words, do you want all processes tohave roughly equivalent performance, or do you want to focus onsome specific processes and make them as efficient as possible?Whenever a single instance of OpenVMS runs on multiple RADs (whetherit is the entire machine, a hard partition, or a Galaxy instance),then you must answer this question, because the answer dictatesa number of configuration and management decisions you need to understand.

The OpenVMS default NUMA mode of operation is "uniform forall." Resources are assigned so that over time each process on thesystem has, on average, roughly the same performance potential.

If "uniform for all" is not what you want, you must understandthe interfaces available to you in order to achieve the more specialized"optimal for a few" or "dedicated" environment. Processes and datacan be assigned to specific resources to give them the highest performancepotential possible.

To further enhance your understanding of the NUMA environment,this chapter discusses the following topics:

skip links to sections within this chapter.
OpenVMS NUMA Awareness
Application Resource Considerations
Batch Job Support for NUMA Resource Affinity Domains
RAD Application Programming Interfaces
RAD System Services Summary Table
RAD DCL Command Summary Table
System Dump Analyzer (SDA) Support for RADs
end of content navigation links


go to previous page: Developing OpenVMS Galaxy ProgramsDeveloping OpenVMS Galaxy Programs
go to next page: OpenVMS NUMA AwarenessOpenVMS NUMA Awareness