What does Guaranteed CPU mean for a VPS Package
Each VPS package is created on a node containing only other VPSs of the same package. i.e. a Xen E5-2620 SM VPS is created on a physical server containing only other Xen E5-2620 SM instances (nodes are homogeneous). As a result of disk space and RAM allocations for a given plan, we are able to put varying numbers of VPS packages on a single node. Nodes that house plans with smaller disk space and RAM allotments have more VPS instances on them. Conversely, nodes that contain plans with large disk space and RAM allotments contain a smaller number of VPS instances on them.
Each VPS is given "fair-share" CPU usage on the server. CPUs sit idle in most cases so, in general, a single VPS is able to use the full amount of CPU resources available to it. Each VPS has access to 4 vCPU and can use the full processing power of 4 vCPU if the CPU resources are available.
Guaranteed CPU refers to the case when all VPS instances on a given node are using all of the their available CPU resources. If the occasion arises that all VPS instances need access to the CPU at the same time, your VPS will be guaranteed the percentage of 4 vCPU listed for that package's "Guaranteed CPU".
For example, the Xen E5-2670 is guaranteed 25% of its 4 vCPU. If, suddenly, all VPS instances on a Xen E5-2670 node needed access to the server's CPU at the same time, each instance will be able to use 25% of the processing power of its 4 vCPU. Otherwise you have access to 100% of the processing power of the 4 vCPU.