How NUMA Allocates Memory (文档 ID 780466.1)

How NUMA Allocates Memory? (文档 ID 780466.1)

转到底部

---

In this Document Goal   Solution   APPLIES TO: Oracle Database - Enterprise Edition - Version 10.1.0.2 to 11.1.0.7 [Release 10.1 to 11.1] Information in this document applies to any platform. Information in this document applies to any platform. Oracle Server - Enterprise Edition - Version: 10.1.0.2 to 11.1.0.7 GOAL How NUMA Allocates Memory? SOLUTION In a NUMA system, processors, memory, and I/O are grouped together into nodes so that each processor is bound to specific memory addresses  By default a NUMA system will choose the local node to allocate memory from and totally exhaust all of the memory from that node before deciding to allocate memory from other remote NUMA nodes.  While this results in holding an object that will fit in within a single NUMA node and avoid fragmentation, it can also result in aggressive swapping on one node while there is plenty of free memory on other nodes. Changing the NUMA memory allocation policy to allow memory allocation from other remote nodes before total memory exhaustion on the local node  will result in objects not properly fitting in a node however it will prevent the exhaustion from one node and the resulting swapping while there is plenty of memory on other nodes.  In Red Hat Enterprise Linux version 3 Update 7. The tunable 'numa_memory_allocator' parameter controls memory allocation policy for x86-64 systems.  To allow memory allocation from remote nodes run the following command as root: echo "1" > /proc/sys/vm/numa_memory_allocator