OpenVMS ApplicationSupport for Resource Affinity Domains (RADs)
The large amount of physical memory in the new AlphaServerGS series systems provides opportunities for extremely large databasesto be completely in memory. The AlphaServer nonuniformmemory access (NUMA) system architecture provides thebandwidth to efficiently access this large amount of memory. NUMAis an attribute of a system in which the access time to any givenphysical memory is not the same for all CPUs.
In OpenVMS Alpha Version 7.2-1H1, OpenVMS engineeringadded NUMA awareness to OpenVMS memory management and process scheduling.This capability (application support for RADs) ensures that applicationsrunning in a single instance of OpenVMS on multiple building blockscan execute as efficiently as possible in a NUMA environment.
The operating system treats the hardware as a set of resourceaffinity domains (RADs). A RAD is a set of hardware components (CPUs,memory, and I/O) with common access characteristics. On AlphaServer GS80/160/320systems, a RAD corresponds to a quad building block (QBB). On AlphaServer ES47/ES80/GS1280systems, a RAD corresponds to a two-processor CPU board. A CPU referencesmemory in the same RAD roughly up to three times faster than itreferences memory in another RAD. Therefore, it is important tokeep the code being executed and the memory being referenced inthe same RAD as much as possible while not giving some processesa consistently unfair advantage. Good location is the key to good performance,but it must be as fair as possible when fairness is important.
The OpenVMS scheduler and the memory management subsystemwork together to achieve the best possible location by:
Assigning each process a preferredor "home" RAD.
Assigning process-private pages from the home RAD'smemory.
Usually scheduling a process on a CPU in its homeRAD.
Replicating operating system read-only code on eachRAD.
HP OpenVMS Alpha Partitioning and Galaxy Guide
Managing Workloads With Partitions
