January 31, 2005 9:00 PM PST
Sun bangs utility computing drum
In November, Sun announced--but didn't actually begin selling--a service by which customers could pay Sun $1 per processor per hour to use raw processing capacity. Now part two is coming, in what the company is calling its Sun Grid: storage costing $1 per gigabyte per month.
The Santa Clara, Calif.-based seller of servers and software believes that computing capacity--though not the computers themselves--is becoming a commodity, and it wants to be the top provider. In its view, customers will cease fussing with their own equipment and simply buy what they need in much the way they buy electricity.
In some ways it's a distant vision, particularly for customers reluctant to entrust their data and computing infrastructure to others, but Salesforce.com, IBM and others are trying to make it reality. It's important for Sun in particular, which is trying to attract new revenue and customers as part of its recovery from the dot-com collapse.
IBM believes it's got the edge over Sun, arguing that its computing capacity on-demand service costs half as much as Sun's. And it has signed partnerships with 60 software companies, including Siebel Systems, to sell their products in conjunction with IBM's utility services.
But some are interested in Sun's service, said Aisling MacRunnels, director of utility marketing. Five major customers in the oil and gas industry and in financial services are planning to join during the next two months, and some of them plan to book a dedicated computing capacity of more than 5,000 processors each, she said.
"We were surprised by the large initial requests. We thought it would take more time to cross that cultural chasm" into the utility computing realm, she said.
That popularity, though, forced Sun to push back the utility computing service for smaller customers, who likely will get access in June, she said.
The storage service, which also is expected to be available at lower prices for those who are willing to accept lower levels of data protection, is scheduled to be available by early April, according to Chief Marketing Officer Anil Gadre.
Three other Sun Grid services are in the works:
An application service, in which Sun inks partnerships with partners who will overlay their software or other products on a Sun infrastructure. Availability is scheduled for this summer.
A developer service, also planned for this summer, where programmers can write software that runs well on a large number of inexpensive networked computers. Most development today is done on single machines, but data centers increasingly are populated by clusters of low-end servers.
A desktop utility, in which Sun cooperates with a high-speed Internet access company such as a cable television company that distributes Sun Ray terminals that connect to Sun back-end servers for computing power. That's scheduled for trials soon, perhaps in Palo Alto, Calif., and commercial availability is scheduled for this summer and holiday season.
The Sun Grid offerings rely on Solaris, the operating system Sun is promoting in favor of rival Linux, Gadre said. Most computing in the Sun Grid will take place on servers using Advanced Micro Devices' Opteron processor.
Nine months ago, Sun launched a different utility storage offering, charging $1.95 per gigabyte per month. But the company decided that method--paying for equipment based on how much it's used--no longer fits its definition of utility computing, MacRunnels said.
The new definition: Utility computing services must be easy to use, available from any location, reliable, and easily and instantly expandable.
Also Tuesday, Sun plans to announce it has aligned its product sales effort to focus on specific markets, Gadre said. Sun now will back up sales representatives with teams specializing in data centers, data management, Web services, desktop computing and mobility, digital identity, and managed services, Gadre said.
And the company is expected to tout an outside testing program for its "Niagara" processor, due to arrive in 2006.
1 commentJoin the conversation! Add your comment