SMP machines preferred in partitioned implementations

Two or more processors run your requests faster than one for a runnable queue of N entries, and four processors are twice as fast, in general. If we cannot segregate the applications with lighter CPU workloads on to a LPAR, then adding more processors to a DLPAR might be more appropriate, because SMPs are inherently less susceptible to the problems associated with mixed CPU load scenarios. The node agent makes this information available to WebSphere XD autonomic controllers. The controllers read that new information and dynamically make use of it in their calculations.


Next