Extracting Resource Allocation Data for Chargeback in a HP Virtual Server Environment for HP Integrity Servers

Car Rental
Mileage
Fuel Level
Duration
Tread ware
Oil level
Transmission fluid level
….
Charge
Asset: Car
CPU
Memory
Storage IO transactions
Storage IO bandwidth
Network IO transactions
Network IO bandwidth
Number of users accounts
Number of processes
….
IT
Charge
workload
Asset: Server
Many usage parameters are available for both cars and servers, but
for charging only a minimum set are typically needed to determine
relative usage of the asset.
Figure 10 Parameter to determine share of cost
In most environments, a suitable primary indicator of a workload’s usage of a physical server is CPU
count. Server cost is generally proportional to CPU count and is sized according to the anticipated
CPU requirements of the workloads that will use it. The remainder of this paper refers to CPU in units
of CPU cores and calls them cores.
A possible secondary indicator of a server’s workload usage is memory utilization. For some
workloads, memory utilization is either a better indicator or is useful as a modifier for determining a
workload’s use of a physical server and its associated cost.
Network and storage usage parameters of a workload relate more directly to WAN and SAN asset
costs – fibre cables, switches, arrays, LUNs, and so on -- and therefore are of lesser importance when
assigning cost of the physical server itself. With that in mind, this paper focuses solely on the physical
server and does not address accounting for workload network and SAN utilization. For those
interested specifically in chargeback in SANs, HP StorageWorks Essentials includes a Chargeback
Manager component for charging back SAN-based storage.
VSE Data for Chargeback Reports
The exact parameters of interest to an IT organization – the form of internal IT chargeback reports and
user-facing chargeback bills – vary based on the services, policies, and VSE technologies being used.
This section describes two possible reporting styles.
Reporting Style 1: IT as Service Provider
This style of report lists all workloads and their level of usage independent of any particular server.
Workload usage is reported in cores or memory, or both, as absolute values; for example, 3.2 cores
and 1.5 GB used. In its simplest form, this information is multiplied by a rate to result in the charge
amount for the owning organization. The resulting amount can be used for any configuration but is
most likely used in environments where:
12