Vendors readily talk the talk in promoting the "value proposition" of server blades: Just add a blade to the chassis and scale to as much processing as you need. But do blades walk the walk in delivering the goods?
"Many users are clearly disappointed in blades' failure to deliver the benefits promised," said Bob Gill, chief research officer of the TheInfoPro (TIP) research firm in an interview with internetnews.com. According to a report just released by TIP, many users that have installed and tested blades report putting widespread deployment on hold.
"The top reasons I heard in my interviews for getting blades were improved manageability, cost, space and energy savings," said Gill. "But those were the same areas they complained about [after installation]."
For example, Gill said he repeatedly heard from blade users that they could not fully populate a cabinet with blades because that density gave off too much heat. Instead, cabinets were filled to about half capacity, thereby undercutting cost and space promises. "If you can't fully populate the chassis, it's like buying an expensive razor and then finding out the blades aren't that cheap," said Gill.
Despite this grim assessment, TIP said roughly two-thirds of those it interviewed envision some value in a blade approach in the longer term (i.e., 2006 and beyond) and have them in their purchase plans. Also, 25 percent of those interviewed said they are aggressively deploying blades that are critical to operations.
TIP said it conducted 70 in-depth, one-on-one interviews with server professionals at Fortune 1000 companies.
Mark Potter, director of HP's BladeSystem division, said HP is investing heavily in the technology and that it's the fastest-growing segment in the company's server business, with 115 percent growth in revenue year-to-year.
HP has delivered three generations of Intel-based blades since 2002 and, more recently, blades based on AMD's 64-bit dual-core Opteron processor.
"Some of our competitors have not thought through all the problems associated with power consumption, and customers have had to upgrade their power supplies three to four times," said Potter. "We've had customers that have deployed competitors' blades and asked us to buy them out in favor of our gear."
Gill said HP, IBM and others have delivered very good technology, but widespread adoption is another year or two away. He compares it to a period in the 1990s where local-area network vendors kept promising the next year would be "The Year of the LAN."
"This is not 'The Year of the Blade,' but it's coming, more like in 2006 or 2007," said Gill. Other key findings in the TIP report:
- Roughly 35 percent of customers currently see blades as unnecessary, costly and an immature technology.
- Roughly 40 percent of interviewees have blades in initial pilots, often at management's request, and often not for a specific or identified need.
- Only 25 percent of users expect to spend more in 2005 than in 2004 on blades, which is low for a supposed growth area. Fully 33 percent actually expect to spend less in 2005 than 2004.
Gill said about 10 percent of the users he interviewed did not adopt blades by choice or to run specific applications. Instead, he said, "In a lot of cases, someone in management said, 'We're going to make a commitment to blades' and gave them to the data center to go figure out what to do with them."
This article was originally published on internetnews.com.