In his presentation at GigaOM’s Structure conference on Wednesday, Feldman explained that the data center is not only the cloud, it’s providing the value for most of the phones, tablets and myriad devices we carry every day.

“The demand for compute has left the client side and moved into the data center,” said Feldman. “Over a three-year period we went from 3 percent to a third of the U.S. population owning a tablet … We now spend hours and hours a day in the cloud where before, we were on the couch.”

This change means we’re not just changing computing, but also networking and storage. He said IT has become software-defined. And the building blocks aren’t the only thing changing, the buildings where we house the compute is changing as well. Even where we put those buildings is changing.

“We used to put data centers in urban environments but where do we put them now? In Eastern Washington or along river banks in Oregon to take advantage of lower power,” said Feldman.

“The data center now does the compute for the client side. Millions of millions of users each with parallel work. We don’t ask it to do CAD/CAM …. the vast majority of the work we ask it to do is simple parallel work for the client side. And that work is very different.”

It’s not about CPU performance, which means that work requires a different type of processor. “And in the future we believe it’s going to be an ARM processor,” Feldman said.

So Feldman called for the industry to rethink how it designs servers to make them more efficient. The server world should also embrace open source hardware like what the Open Compute Project wants to offer. He left the audience with the thought that in the 60 year history of computing, smaller, higher-volume parts have always won. That used to be x86-based processors. But in 2012 more than 8 billion ARM CPUs were shipped, more than twenty times the x86 volume.

It’s too bad Feldman didn’t spend some time talking about how AMD plans to adapt to the realities of an ARM-based chip world, where dozens of vendors have the ability to design and build ARM-based chips. That’s a big shift from building x86 chips that only two vendors can sell.

Check out the rest of our Structure 2013 coverage here, and a video embed of the session follows below:

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• The role of converged infrastructure in the data center
• How tomorrow’s mobile-centric data centers will look
• Webscale and cloud are changing the server value chain. Who wins?

Dell increased its server revenue and shipments in the first quarter, according to new Gartner and IDC figures, while archrival Hewlett-Packard’s server share declined. Gartner said Dell boosted server revenue by 14.4 percent year over year, and IDC put revenue growth at 10.1 percent. The problem is, lesser known server vendors edged out Dell in both categories.

After all, while its market share in shipments, at 22.2 percent, has gone up slightly since 2006, when it was at 21.7 percent, the other guys have jumped forward in a big way, from 27.6 percent to 37.5 percent. Going private, which is still in the works, could let Dell change up its operations without having to deal with shareholder scrutiny on profitability and other matters.

If Dell can execute on the plan quickly, it could gain advantage as other legacy vendors such as Fujitsu and IBM scramble to catch up and lower prices to stay competitive with Quanta, Wistron and other “no-name” vendors that have seen their market share rise. But those “other” vendors are shrewd and have sold boatloads of gear by building custom servers for a new wave of webscale customers. Quanta and their kind have figured out ways to keep costs low at scale — their shipments are more impressive than their revenue.

That’s why Dell might gain ground with this initiative against name-brand server makers, but it will have to try harder to fend off the little guys.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How fourth-quarter 2012 will affect IT spending in 2013
• A near-term outlook for big data
• Will Storage Go the Way of The Server?

According to IDC,  revenue for every name-brand server maker except Dell and Cisco, was down. Market leader Hewlett-Packard saw its revenue for the period off 14.8 percent year over year. (On the last HP earnings call, CEO Meg Whitman acknowledged that the company is passing on low-margin server deals.) Looks like Dell, with revenue up 10.1 percent year over year, is picking up the slack.

Cisco, which is building from a much smaller base, was up a whopping 34.9 percent, but it’s still got a small slice of the pie with Q1 revenue at $450 million, compared to $2.9 billion for HP, $2.8 billion for IBM and just over $2 billion for Dell. Cisco’s market share rose to 4.1 percent from 2.8 percent.

Oracle’s hardware struggles continue — its revenue was off 26.2 percent year over year and its market share fell to 4.8 percent from 6 percent last year.

Update: Whoa, here are some other eyebrow-raising stats, thanks to IDC Group VP Matthew Eastwood:

Some of the companies with smaller share numbers, saw phenomenal growth in units shipped year over year. Lenovo, for example,shipped 48.8 percent more servers this year over last, but that pales in comparison to SGI (SGI??!!) unit growth which was up 98.7 percent. Chinese giant Huawei, however, saw the steepest growth curve, with units shipped up a whopping 156.6 percent. Huawei saw its overall market share raise from 0.6 percent to 1.6 percent.

But most interesting is that unit shipments from “other” companies were actually off 2 percent year to year. Didn’t expect that.

The much-watched non-brand-name “other” segment — which includes companies like Wistron and Quanta — saw aggregate revenue fall 1.7 percent but market share growing to 14.9 percent from 14 percent. Companies running huge data farms — the Googles and Facebooks — often turn to these companies for white box servers built to their specifications, a development that has to irk the name brands who seem to be seeing the addressable market for their servers shrinking.

Here’s Jordan Novet’s take on the Gartner research.

And now for the IDC numbers: 

This story was updated at 6:18 a.m. PDT with additional information on unit shipments from IDC and again at 8:38 p.m. PDT to correct the Oracle server market share figures in the text.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How fourth-quarter 2012 will affect IT spending in 2013
• Takeaways from the second quarter in cloud and data
• A near-term outlook for big data

In the first quarter of 2011, “other” vendors claimed 32.1 percent of marketshare in terms of unit shipments. HP had almost 30 percent, and Dell had 22 percent. Now the “other” guys in aggregate have moved forward and hold 37.5 percent of the market in terms of units shipped; HP has dropped to 24.9 percent, while Dell’s share was roughly flat. IBM in the same period fell from 11.8 percent to 9.9 percent.

Source: Garnter

HP is trying to shake things up with its new energy-efficient Moonshot servers. The microserver bet looks like a good one, although Moonshot’s adoption is an open question. An HP executive told my colleague Barb Darrow that the data showing the growth of white-box vendors emerged “before the world knew about Project Moonshot.”

Dell, for its part, grew year over year last quarter, but far more modestly than the “other” vendors. It’s unclear what going private will mean for the company in terms of its server sales.

Cisco grew, too, but its chunk of marketshare — 2.3 percent — is small, so growth appears larger than it actually is.

Oracle is absent from the list of vendors measured by server shipment, although its server business doesn’t look promising, either, at least not as much as for the “other” vendors. Oracle server revenue slipped more than 27 percent year on year.

The growth of Facebook, Google and other webscale companies out of colocation facilities and into their own data centers has introduced the opportunity to go with tailor-made gear that make sense at economies of scale. (At GigaOM’s Structure conference in San Francisco in three weeks, this topic might very well come up in conversation with Werner Vogels, chief technology officer at Amazon.com, and Jay Parikh, Facebook’s vice president of infrastructure engineering.)

Some companies running lots of applications have moved in this direction as well — take Amazon and Rackspace, for example. Hence the rise of companies that do this for webscale players.

With more hardware designs becoming available through the Open Compute Project, still more business could flow to no-namers in the years to come. And that’s to say nothing of switches, a piece of hardware certainly ripe for commoditization.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How tomorrow’s mobile-centric data centers will look
• Migrating media applications to the private cloud: best practices for businesses
• How fourth-quarter 2012 will affect IT spending in 2013

The thought to write about what “best” means in technology came to me after reading Joe Weinman’s book “Cloudonomics”. In the book he points out several times that having the best technology doesn’t guarantee that you’ll end up with the best solution or service. So how do you determine what’s best?

Some definitions of “best” in IT include:

• Most comprehensive solution
• Lowest priced solution
• Easiest solution to install and get up and running
• Best performing in high latency situations
• Doesn’t require capital
• Doesn’t lock me in
• Highest price

I’m guessing that after having read the seven bullets above, you’re already starting to get a sense of how we sometimes make assumptions about the appropriateness of an IT solution based on incomplete considerations. It seems self-evident what best should mean, but it’s often true that we don’t recognize how priorities can contradict or shift our definitions.

Buying the best

When you buy the best solution in any product category you assume that you are in fact getting the most appropriate solution for the money. That value might be any combination of things from a name (such as Hermes,) to performance or sex appeal (like Ferrari,) or maybe time per transaction (I won’t wait in line at Whole Foods markets).

The reality is that best could mean all of those things or none of them. While it’s almost certain that a bag you buy from Hermes is going to be well made, is it 30 times better than another brand? Is a $1,000 bottle of wine 20 times better than a $50 bottle? If money is no object to you, then the answer is more likely to be yes, a $1,000 bottle is better. On the other hand, if you value the ability to have different bags for different events, or simply prefer the ability to buy a new bag more often, then the $150 bag is probably “best” compared to the Hermes bag.

Buying IT is no different, except that it’s immensely more complex. In IT there are myriad variables that affect the ability to get the most from any solution. These variables include price, features, latency, maintenance, flexibility, open vs. proprietary, required training, user interface, APIs, and more. What about your team’s ability to sell the solution as the right choice to your customers? How about whether or not you’ve got the correct organizational and financial structure to support the solution appropriately?

How to determine what’s best

The decision process for every new technology or solution selection needs to cover a wide list of criteria. These criteria will mostly all be the same for each business but the priorities will change depending on your organization.

The majority of the standard selection assumptions (need, ROI, cost, etc) are well understood, but even among those “standards” there is room for better decision making and prioritization. I like to include the following non-standard criteria when my team is making a solution selection.

• How much value do we get out of the solution at 80 percent of total feature set?
• What other capabilities does this solution open the door to in the future?
• How many of my customers need to use the solution and to what extent before it adds new value?
• What organizational support (business and IT) do I have for the long term “ownership” needs (staff, training, champions, budget, lifecycle, etc.)?
• How does this solution position my team to execute against larger IT and business visions?
• Does this solution leverage other partners and technologies already in use?
• What’s the time to install vs. cost to purchase or time to benefit? (In other words, will I get 30 percent net new benefit value in year one vs. nothing in year one and two, but 80 percent in year three from another solution?)
• Ownership risk assumptions (what assumptions are you making at the front end of any solution selection and are those assumptions still accurate? With modern cloud/SaaS etc., you might not have the ownership risk you “enjoyed” with many legacy platforms)

While I could easily argue that all of the above bullet points are important to every organization, they must each be measured against the organization’s situation at the time. Are you low on cash, but growing fast, do you have a higher risk profile or regulatory concern? The process of prioritization can only be done by the business making the purchase.

Seems like an oxymoron

Sometimes the best purchase is the purchase you avoid, other times the best purchase is the one you didn’t make. IT is littered with examples of purchases that would have been better left undone. However, just as common are those purchases that were never made because they weren’t “perfect”. When you’re looking for perfect, keep in mind that there’s no such thing in software and many cases in hardware, but if you can solve a problem even at 70 or 80 percent, the purchase might still be better than waiting for the “perfect” option.

The test and fail option is much more real today than it ever has been and it’s a good thing. Now you can test, fail, and retry three or four times all for less effort and cost than making one selection in the past. So, step forward boldly, but don’t forget that when you are thinking “best” make sure you’ve really developed a case for what best means to your organization.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Quality of the cloud: best practices for ISVs
• Infrastructure Q1: Cloud and big data woo enterprises
• Forecasting the future cloud computing market

Interest in liquid cooling — as opposed to standard air cooling — in applications other than high-performance computing has been slowly rising since Austin, Texas-based Green Revolution and another provider, Iceotope, came out of stealth mode in 2009, but Green Revolution in particular has seen a tidal wave of inquiries in the past nine months or so, Best said.

Google has shown interest; the company got a shout-out from Amazon Web Services Distinguished Engineer James Hamilton and Intel also liked the results of a year-long test of servers inside the Green Revolution gear. Since the Intel news, there were “a couple big people who started testing us, and those people have been talking,” Best said.

Enterprises are more risk-averse than webscale companies and don’t care as much about cost savings, but Best said he thinks wider enterprise adoption could be just three to five years away.

The question is whether more data center administrators will be able to wash their hands of concerns about removing fans from servers, bringing in specialty racks and — not to mention — splattering oil on themselves. GigaOM Research Analyst Pedro Hernandez pointed out these issues (subscription required) in late 2009.

But webscale companies can skip the process of modifying servers to fit liquid-cooled racks and just buy custom servers through legacy vendors or lesser known manufacturers with original-design roots that are emerging as direct sellers, such as Quanta. And with more webscale companies rolling up their sleeves, Best’s enterprise forecast isn’t so hard to believe.

This story was corrected at 4:25 p.m. PT to remove references to the need to seal hard drives when using the Green Revolution Cooling system.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How fourth-quarter 2012 will affect IT spending in 2013
• The role of converged infrastructure in the data center
• How tomorrow’s mobile-centric data centers will look

CloudFlare Founder and CEO Matthew Prince detailed the problems the company is trying to solve in a blog post earlier this week. In a nutshell, although its network edge that spans 23 data centers is (usually) capable of handling most traditional DDoS attacks, there are a couple types of attacks that target different bottlenecks at the local area network level. In these cases, the 1 Gbps networks ports on CloudFlare’s servers can get overwhelmed, as can the processors themselves.

Of course, when you’re running a multitenant cloud-based service like CloudFlare is, these types of events take on a different urgency:

“Both these problems are annoying if it affects the customer under attack, but it is unacceptable it spills over and affects customers who are not under attack. To ensure that would never happen, we needed to find a way to both increase network capacity and ensure that customer attacks were isolated from one another.”

So, over the course of 2012, CloudFlare spent its time working on what it calls “Project Bondage.” Essentially, that meant configuring the individual ports to look and act like a single port capable of handling much more bandwidth, and then reworking the CloudFlare operating system to prevent external CPU-level attacks from affecting internal workloads.

But the company didn’t stop there. Prince wrote in the blog that CloudFlare’s next-generation servers feature 10 Gbps ports to significantly increase network bandwidth even without port bonding. In an email, he confirmed that rather than use off-the-shelf servers as it has been doing, CloudFlare’s “G4″ servers were designed in tandem with and built by Quanta, the same company that builds Facebook’s servers as well as servers for other large web companies.

CloudFlare still uses off-the-shelf Juniper switches but, Prince added, “[W]e’re tinkering.”

Feature image courtesy of Shutterstock user teflon_timmy.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How tomorrow’s mobile-centric data centers will look
• Web startups: How to guard against security breaches
• What the Data Center World Can — and Can’t — Take from Facebook’s Open Compute Project

Reports surfaced last week that Hurd, who is co-president of Oracle, was on the short list of prospective Dell CEOs from Blackstone Partners, a private equity firm interested in buying Dell. Blackstone, presumably, would need to install its own guy if its bid against the current Dell guy, Michael Dell, wins the day.

Asked about the issue in Japan at a press event, Hurd said: “I’m very happy at Oracle. No interest.”

Mark Hurd in Japan goes on the record on #Dell: “I’m very happy at Oracle. No interest." #oracle http://t.co/pYsKAM9fDK—
deborah hellinger (@dhellinger) March 25, 2013

Of course, that doesn’t mean Blackstone didn’t ask.

The timing must have been awkward for Hurd. Later today, his boss, Oracle CEO Larry Ellison, and systems guy John Fowler will unveil the latest-and-greatest Oracle server. If you want to sit in, you can register here. 

It doesn’t help when a humongous software company is trying to build hardware credibility for there to be rumors about one of your top guys — a hardware guy — jumping ship to another hardware company. Hurd is the former chairman and CEO of Hewlett-Packard who exited under a cloud in August, 2010.  A month later he was  snapped up by Ellison as Oracle co-president (with Safra Catz)

First line of Blackstone/Michael Dell meeting: "So, yeah… sorry for calling Mark Hurd before we called you."—
  (@danprimack) March 26, 2013

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How fourth-quarter 2012 will affect IT spending in 2013
• Will Storage Go the Way of The Server?
• IT spending update, third quarter 2012

It all started, Mike Yang says, with a conversation he had with Facebook’s vice president of technical operations in 2007 or 2008. Rather than source servers through a traditional vendor like IBM for its data centers, Facebook turned to Quanta.

Back then, Quanta didn’t sell servers directly to customers, it only built them for traditional server vendors who then put their name on them and sold them to customers. Fast forward a few years, and a majority of Quanta’s server revenue stems from direct deals — 65 percent in 2012, and a forecasted 85 percent this year. Now, it counts other large-scale server buyers such as Rackspace among its customers.

Mike Yang. Source: Quanta

Yang, the man in charge of Quanta’s cloud computing business unit, beamed during an interview on Thursday as he spoke about how the company can directly offer energy-efficient and high-performance products for webscale customers and smaller ones, too. If the Taiwan-based hardware maker’s 85 percent forecast proves out, the company could become a more recognized supplier for cloud computing venues, further threatening old-line server vendors like Hewlett-Packard and Dell.

The company, with U.S. headquarters in Fremont, Calif., didn’t show projections of server revenues in dollars or server shipments in total but said it shipped 1.2 million server motherboards in 2012 and plans to ship at least 10 percent more — 1.32 million motherboards — this year.

Quanta appears to be on a roll with Quanta-brand direct server sales growth. At the same time as it’s doing custom jobs for webscale customers, it’s also promoting direct sales of other gear, including off-the-shelf storage and network appliances, to smaller customers through a subsidiary Quanta established last year, Quanta QCT.

The company has a few strategies in mind for shifting from an original-design manufacturer to a name brand in its own right, at least in servers. It sees full racks of equipment, under the Rackgo name, as a major seller this year. The Rackgo offering, which includes compute, storage and network appliances, can appeal to customers because there’s simply one company to go to when problems arise, Yang said.

And then, of course, there’s the Open Compute Project — the Facebook-led open-source hardware initiative that kicked off Quanta’s evolution as a direct server vendor. Quanta will come out with multiple products based on Open Compute specifications later this year, although exact timelines weren’t immediately available.

Next month, the company will open an office in Seattle in order to be closer to customers. Yang said Quanta has several customers in the area, although he declined to name them. Microsoft, which is building huge data center capacity for Windows Azure and its Live offerings, is a short drive from Seattle, in Redmond, Wash., and Seattle is much closer to Quincy, Wash., a hotbed of data centers, than the Fremont office. Quanta will add more U.S. offices for sales and service this year, Yang said.

Quanta is also opening up to the press, rather than silently working behind the scenes. That campaign started last year.

The company’s business model has undergone a sea change. If the upward trajectory keeps up and the server-market dynamics keep shifting in its favor, Quanta could become one of the stalwart name brands of IT technology.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How fourth-quarter 2012 will affect IT spending in 2013
• The role of converged infrastructure in the data center
• A near-term outlook for big data

Those optimistic about 2013 point to the second half of the year (now only four months away) as when we’ll begin to see some stronger growth. In their view, Europe will be better, the U.S. will be stronger, China will begin to reaccelerate in earnest, currency will be stable, and (perhaps most importantly) the year-over-year (YoY) comparisons between 2012 and 2013 will be easier because mid-2012 was so weak. Moreover, as EMC CEO Joe Tucci stated during the company’s recent earnings call, “You can’t starve IT for too long,” which suggests that better times must get rolling again. On the other hand, CEO John Chambers’ statement in the Cisco earnings call, that perhaps 2012 represents the new normal, might be a better indicator of what we can expect in 2013. We’ll see.

That said, here’s what we’re expecting for 2013.

Table 1. 2012 and 2013 worldwide IT spending (in $ billions)

Overall, smartphones and tablets will lead the way, with a 5.6 percent worldwide IT growth rate in 2013. While Apple’s tepid forward guidance in its recent earnings call and its implications for smartphones and tablets gives us pause, we expect the category to continue to lead, with growth in the 20.4 percent range in 2013. In hardware, 2013 will look like 2012, with smartphones and tablets buoying the otherwise-weak spending in the PC, peripheral, and server segments. (This will be offset somewhat by stronger storage and network hardware spending.) Software continues to flourish — especially solutions-related software like SaaS-enabled customer-relationship management (CRM), supply-chain management (SCM), and industry vertical systems like health care. This sector will continue to grow in 2013. Finally, services has experienced a punishing 2012, and this area will only look modestly better in 2013.

A methodological note: We’ve reevaluated our model and determined that we have underweighted business-process outsourcing (BPO) as a service category. As such, we are restating 2012 and 2013 growth rates to accommodate a higher weighting of BPO in our services spending. The net is that growth in services and all spending are a bit more robust, since BPO is growing faster than the aggregate of traditional IT services. The full methodology can be found at the end of this report.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• IT spending update, third quarter 2012
• IT spending update, fourth quarter 2012
• Work media tools in 2012 and beyond