Tessellation strikes a balance between maximizing resource utilization to achieve performance and selectively idling resources to provide QoS guarantees. Decisionmaking logic is packaged into a Policy Service that distributes resources to Cells by combining system-wide
goals, resource constraints, and performance targets with
current performance measurements (see Figure 3). The
results are passed as a Space-Time Resource Graph
(STRG) to the Tessellation kernel for QoS enforcement.