Palo Alto, California-based HP’s pay-per-use service is delivered through its Financial Services organization, and sees the adoption of automated metering technology that records the actual utilization of each CPU deployed within the Superdome server and charges users based on the amount of processing power used.
This has advantages over existing capacity on demand systems (where servers are equipped with more processors than are immediately required in order to cope with peaks in demand), in that customers can still have additional processing power available for peak periods, but do not have to pay for spare processing power when it is not required.
The metering technology works by automatically collecting CPU utilization information, encrypting it and securely transmitting it to HP’s billing engine, according to the company, while customers will be billed once a month for their actual use of processing power.
Capacity on demand servers and pricing models were first introduced by the major Unix server vendors in November 1999 as first Sun Microsystems Inc, then HP, and finally IBM Corp introduced their processing on-demand schemes. HP can now claim to be the first vendor to have pay-per-use capacity metering on Unix servers generally available, however.
Earlier this week, IBM outlined forthcoming autonomic computing tools that will enable it to better deliver capacity upgrade and downgrade on demand capabilities, which are currently being offered to customers through customized service deals. If the race to get capacity on demand packages out of the door is anything to go by, expect similar automated metering schemes from IBM and Sun to hit the streets soon.
Source: Computerwire
