Sun's fight for the greener data center
Nicholas Carr, author of the book, "Does IT Matter?, critiques Sun Microsystems' marketing as represented by Jonathan Schwartz's blog, claiming that "Sun has bounced between marketing pitches like Ricochet Rabbit on a meth jag". But then he latches onto Sun's latest initiative on power-efficient data centers and servers, and notes that that is a trend worth going after, given many of Google's and Intel's recently voiced concerns over the same topic.
I don't know. This sounds a bit to me like SUV drivers suddenly waking up and deciding if gas is going to be four dollars a gallon, hybrid technology is a good idea. After all, the fundamental trend behind "Does IT Matter" was the commoditization of information technology. The claim was always that since any company could buy racks of Dell servers, there was no competitive advantage in those servers. But those are the same servers that Sun customers are trying to displace. So which is it? Is IT commoditizing (and therefore Sun's power saving approach doesn't matter, since they are simply competing against a commodity), or has the business world finally recognized that Sun's approach of designing its own chips makes some sense after all, since it is those chips that provide better power consumption (and thereby, greener data centers)?
And by the way, much of the need for gazillions of servers in data centers has been from pragmatic approaches to getting Windows servers to run reliably. About 10 years ago, data center managers discovered that if you want an application to run predictably on a Windows server, you should run it as the ONLY application on that server. So when companies decide to run 300 business applications and databases, they end up with, you guessed it, 300 Wintel servers, each of which is mostly idling and burning power. Sun always claimed they had a better way -- buy a big 64-way Star Fire server, and run as many apps as you like, since Unix doesn't have the same reliability limitations as Windows. But businesses never liked the million dollar price tag of those servers. Instead, they bought 300 $10,000 Wintel servers -- and spent three times as much as they would have for the Sun box.
My view: power consumption is an important and often-overlooked metric. Just as businesses need to consider total cost of ownership (TCO) in their purchases of PCs, businesses also need to consider data center costs of floor space, air conditioning, power, and flexibility in their choices of server technology. Call it total costs of service or TCS. Smart companies will sell solutions with lower TCS, and that's exactly what Sun is doing with their new Star Fire servers. Companies that aren't so smart will tell you that it's all a commodity, and that you should buy their commodity because it has a lower price. Professor Carr is discovering that Sun's marketing and products are different -- but you have to look beyond the commodity IT market to appreciate its value. Let's hope more businesses figure this out.
I don't know. This sounds a bit to me like SUV drivers suddenly waking up and deciding if gas is going to be four dollars a gallon, hybrid technology is a good idea. After all, the fundamental trend behind "Does IT Matter" was the commoditization of information technology. The claim was always that since any company could buy racks of Dell servers, there was no competitive advantage in those servers. But those are the same servers that Sun customers are trying to displace. So which is it? Is IT commoditizing (and therefore Sun's power saving approach doesn't matter, since they are simply competing against a commodity), or has the business world finally recognized that Sun's approach of designing its own chips makes some sense after all, since it is those chips that provide better power consumption (and thereby, greener data centers)?
And by the way, much of the need for gazillions of servers in data centers has been from pragmatic approaches to getting Windows servers to run reliably. About 10 years ago, data center managers discovered that if you want an application to run predictably on a Windows server, you should run it as the ONLY application on that server. So when companies decide to run 300 business applications and databases, they end up with, you guessed it, 300 Wintel servers, each of which is mostly idling and burning power. Sun always claimed they had a better way -- buy a big 64-way Star Fire server, and run as many apps as you like, since Unix doesn't have the same reliability limitations as Windows. But businesses never liked the million dollar price tag of those servers. Instead, they bought 300 $10,000 Wintel servers -- and spent three times as much as they would have for the Sun box.
My view: power consumption is an important and often-overlooked metric. Just as businesses need to consider total cost of ownership (TCO) in their purchases of PCs, businesses also need to consider data center costs of floor space, air conditioning, power, and flexibility in their choices of server technology. Call it total costs of service or TCS. Smart companies will sell solutions with lower TCS, and that's exactly what Sun is doing with their new Star Fire servers. Companies that aren't so smart will tell you that it's all a commodity, and that you should buy their commodity because it has a lower price. Professor Carr is discovering that Sun's marketing and products are different -- but you have to look beyond the commodity IT market to appreciate its value. Let's hope more businesses figure this out.
Technorati Tags: Advertising, Communication, Marketing, Processors, Servers, Star Fire, Sun
posted by Carl Howe at
7:47 PM
| View blog reactions
| 
