You can view details about the nodes on your server using the sys.dm_os_memory_nodes DMV. You will always have at least one memory node, which has a memory_node_id of 0.You will have more than one memory node if you have configured Non-Uniform
Memory Access (NUMA) or if your system uses AMD processors and you have more than four CPU cores.
For NUMA
A non-uniform memory access (NUMA) architecture refers to a system where there the total available memory is broken down into multiple nodes, with groups of processors connecting directly to each node of memory, which is considered the "local" memory pool.
Each node also has access to all other memory on the other nodes, which are considered remote or nonlocal memory. NUMA systems are often referred to as ccNUMA, which means cache-coherent NUMA. The cache coherent (cc) part of this term reflects the fact that
additional logic is built into the system to ensure that any piece of data is only ever loaded into cache at one place at any point in time.
A NUMA system architecture has the advantage that each processor in a node doesn't pay the cost of using a shared bus to access memory, provided the data it wants is in the local memory pool. If the data it wants is in remote memory, then the cost of getting
access to it is a little higher than on an SMP system. Therefore, one of the objectives with a NUMA system is to try to maximize the amount of data that is used from the local memory, and avoid accessing data that is in remote memory.
This text is from Book:Professional SQL server 2008 internals and troubleshooting..please get a copy of this to read in more details