Introduce Last Level Cache (LLC) as a resource with Resource Management Daemon and Resource Director Technology

Nova currently does not have mechanism to use Last Level Cache (LLC) as a schedulable resource similar to CPU or Memory. Intel Resource Director Technology (RDT) enables controlling LLC per workload basis by means of Cache Allocation Technolgoy (CAT). Cache associations per workload are controlled via associating workload with construct called cache ways that divide LLC on each socket into equal chunks of fixed size and constructing a Cache Bit Mask (CBM) called Class of Service (COS). Although the application of CAT is fairly simple scaling the workloads with the limited amount of LLC per socket on a host platform needs to consider multiple aspects to meet the required Quality of Service (QoS) like assigning isolated versus overlapping COS for workloads, dividing LLC in to tiered classes for varying QoS, dynamically construct CBMs at schedule time, among others. For scheduling purposes Nova needs a simple way to know cache availability at schedule time, mechanism to address various QoS and method to assign isolated versus overlapping cache ways.

We propose to update scheduler code to interact with Resource Management Daemon, open source software that communicates using REST API, to provide cache availability using hospitality score for new workloads based on current workload occupancy and mechanism to provide various QoS policies while exposing cache as a resource.