The memory node of the memory node is divided into several pieces of memory. In the NUMA system, a NUMA node corresponds to a memory node. The main function is to provide the allocation location. This statement comes from the SQL Server survival tips that are hard to understand. It should be used to allocate memory. It can also be understood as the original location of the physical memory.
The memory clerk memory. If the Memory Manager client needs to allocate the memory, it must first establish the memory clerk. DMV
2008: returns the set of all active memory clers in the SQL server instance.
2012: returns a set of all active memory splitters in the SQL server instance. Why should we translate it into a memory clerk? how nice it is to translate it into a distributor. The memory object of the memory object is a heap. The memory is allocated by requesting the memory clerk. There are three types of memory objects. 1. variable memory objects 2. incremental Memory Object 3. cache of fixed-size memory objects in SQL Server is mainly divided into two types: Buffer Pool and cache buffer pool in high-speed process. The buffer pool consists of 8 K fixed-size distributors and page caches. all explained, next, let's talk about the figure. The tip of this figure is the Memory Manager component and its relationship. But it seems to me like the memory allocation process. One memory object can obtain the memory from the buffer pool through the page distributor, or from the memory clerk. But after all, the memory of the two is for the memory node. We can see that the actual memory is installed. The memory allocation only assigns the pointer of the physical memory to a variable.