You can't be too rich, too thin or have too much computing power, but you can spend too much money. Since the first server went into a rack, enterprise IT departments have been balancing cost and performance in the data center. HP's ProLiant BL660c Gen8 blade server and DL560 Gen8 server give them new ways to stack the odds. 

The single largest operating cost in most data centers is power, and while estimates vary, as much as 50% of that power goes for cooling. Big data, social media and the proliferation of cloud-based applications and mobile clients have spiked demand for server-side processing and throughput. As the installed base grows, so does the cost of energy, further straining IT budgets.

In response, HP is leading the server industry's focus on server efficiency. By optimizing power consumption and increasing server density, data centers can significantly reduce their energy needs. Density coupled with increased processing capabilities provides a more compact management environment for administrators, while producing a much greater compute output for a given floor size.

HP's latest additions to the ProLiant Scale-up (four or more sockets) portfolio have taken the next steps toward addressing power, performance and floor space congestion. Featuring Intel® Xeon® E5-4600 processors, the BL660c Gen8 and DL560 Gen8 were built to deliver industry-leading performance for compute-intensive applications, while reducing the overall infrastructure and management burden. The results are impressive.

Savings

The most noticeable features of both the BL660c and the DL560 are embedded intelligence and size. While they offer the computing power of many 4s/4U servers, the DL560 is just 2U high. The BL660c is a full-height, single-slot blade. By increasing server density and enhancing energy efficiency, data centers can reduce their overall rack space profile, cooling requirements and power consumption. As a result, HP estimates break-even times of as fast as three months.

HP has built in a number of embedded technologies to automate these savings. For example, Automated Energy Optimization builds a heat map of the entire server, boosting cooling resources where they're needed and slowing them where they aren't.

Manageability

Human resources have a cost, too. Optimizing administrative tasks can reduce the amount of time IT spends on routine management - freeing staff to create more value through innovation.

The new ProLiant servers score here as well. HP's ProActive Insight Architecture provides monitoring and automation tools that can reduce the administration workload by up to 69% by assisting provisioning, monitoring, diagnostics and support, thus allowing IT staff to better support corporate business strategies. 

From the hardware up, the BL660c and DL560 were built for serviceability. All four CPU sockets and memory modules are placed on the system board for easy access, and the FlexibleLOM card slot replaces the embedded LAN on the motherboard to give administrators a wide variety of options for fabric, port quantity and chipset. Additionally, the DL560 provides up to six full-height card slots (two of them full-length) for maximum flexibility. There's even support for 150 Watt cards.

HP's Integrated Lights Out (iLO)

Performance

Savings and efficiency don't matter much if the server can't perform. The ProLiant BL660c and DL560's balanced architecture enable them to command top spots in all cases. Leading SPECjbb2013 and SAP SD 2-tier benchmarks demonstrate the servers' suitability for business processing implementations - including databases, Enterprise Resource Planning (ERP) and batch processing - and the combination of the industry's top floating-point calculations and super-fast PCIe Gen3 I/O provide more than enough horsepower and bandwidth to handle High Performance Computing (HPC) jobs – particularly as control nodes. Finally, in the omnipresent virtualization space, the ProLiant BL660c and DL560 hold the top spots for dense 4-socket servers. High scores across-the-board show that the ProLiants are solid, well-built machines - not just generic servers "cooked" to score well on a specific test.

The Optimum Workloads

Dense 4-socket (and greater) servers like the BL660c and DL560 really show their value in situations that require fast I/O, heavy threading and large amounts of memory. Traditional IT infrastructure operations are often fairly low-demand, so unless you're operating at an incredibly large volume or plan to grow quickly, you may be able to get by with lower-cost, 2-socket servers, like the HP ProLiant BL460c Gen8 server blade or DL360p Gen8 rack server.

Four-socket servers are best-suited to applications demanding large numbers of cores, database threading, or high numbers of VMs.

As you step up your processing load and number of connections, though, the BL660c and DL560 come into their own. Virtualization, business intelligence and business processing are full of potential bottlenecks that won't slow them down. These Gen8 servers are ideal for replacing aging RISC/Unix systems as well as demanding database workloads - either bare-metal or virtualized.

Much more than some Frankenstein's monster artificially packing in resources to make the form factor more dense, the BL660c and DL560 Gen8 servers deliver all this in a well-architected, superbly built package without compromising on performance, scalability or efficiency.

Intel, the Intel logo, Xeon, and Xeon Inside are trademarks or registered trademarks of Intel Corporation in the U.S. and/or other countries.

Facebook is planning to build a massive data center in Altoona, Iowa, the company said on Tuesday. That's right, Altoona, Iowa, a suburb of Des Moines.

With more than a billion users around the world to support and just three wholly owned data centers (Forest City, North Carolina; Prineville, Oregon; Luleå, Sweden, with the latter two still being built out) Facebook may have needed another location. (The company has also stashed servers in at least two co-location facilities owned by other companies, on both the East and West Coasts.) But why Altoona, Iowa?   

Why Iowa?

According to The Des Moines Register, which deserves credit for breaking the story on Monday, Altoona officials sold Facebook on four key selling points:

1. The site sits on the nexus of an interstate fiber optic system, providing connectivity to the rest of the nation.
2. A power substation sits within half a mile of the campus.
3. Transportation access.
4. Environmental stability.

The last is an increasingly important consideration. Data-center providers that went down during Superstorm Sandy in New York last year learned that lesson well; hurricanes, earthquakes, tornadoes and other natural disasters can bring a cloud services down just as effectively as a power outage. 

A Facebook blog post, meanwhile, cited "an abundance of wind-generated power" as well as proximity to "a great talent pool that will help build and operate the facility" as reasons for building in Altoona. Apparently, Des Moines and Ames are the new Silicon Valley and Boston when it comes to technical skills. The new facility will break ground this summer and begin serving traffic in 2014, Facebook said. According to the Register, Facebook's facility "will join what’s becoming a data center corridor of sorts in Altoona. LightEdge was built in 2006, and Enseva will break ground this spring."

Facebook hasn't confirmed the size of its new data center, but the Register earlier this month claimed that planning documents put it at 1.4 million square feet and said Monday the total investment could hit $1.5 billion. That's about four times the size of the company's Prineville facility - and 50% larger than Apple's $1 billion investment in its new data center in Maiden, North Carolina. 

Apple's data center in Maiden, N.C. (Source: Apple)

"In the coming years, as our service continues to grow and people share and connect in more ways, we need to make sure that our technical infrastructure also continues to scale," Facebook's Jay Parikh said in the blog post. "Our goal is not just to deliver you a fast, reliable experience on Facebook every day – we also want to help make connectivity a universal opportunity. Our data centers are essential for making that happen."

How Facebook "Hacks" Its Data Centers

Facebook has put almost as much technology effort into its data centers as its core services. Earlier this year, Facebook disclosed that its Luleå facility would be entirely built on hardware constructed by no-name server manufacturers using designs developed by the Open Compute Project, which shuns "vanity" hardware sold by traditional server vendors like Dell and Hewlett-Packard in an effort to minimize cost. Rather than pay top dollar for the most sophisticated and powerful equipment, this kind of "open source hardware" approach adds capacity by just adding ever more cheap, generic servers.

(See also Can Servers Save PC Manufacturers? Sadly, No.)

Facebook's Lulea, Sweden data center. (Source: Facebook)

Facebook also has been a pioneer in using natural or ambient cooling its data centers. Traditionally, data centers place servers on raised floors cooled by mechanical "chillers," or air conditioners, that push away heat from the servers to keep them running properly. 

Facebook's Prineville facility uses a combination of evaporated water and ambient air to cool the servers without the need for energy-hogging chillers; its Swedish site uses the frigid near-Arctic air to do the same thing. (Google, meanwhile, is building a data center in Hamina, Finland, which pumps water - and exchanges heat - from a nearby canal.) Although Facebook hasn't disclosed how its Altoona servers will be cooled, it's likely to employ some form of evaporative cooling.

Last week, Facebook was the first to offer a near-real-time look at the Power Usage Effectiveness (PUE) — the all-important batting average of a data center's energy efficiency  — of both its Prineville and Forest City facilities. A few years ago, a PUE of 1.8 was considered average; the Prineville facility's PUE now regularly pushes below 1.10, close to the 1.0 ideal.

Lead image via Facebook.

Who is the most successful server manufacturer today? Viewed one way, the answer is "no one". And that face is a dismal warning to traditional PC makers who are counting on servers to keep their businesses afloat as the PC market slowly disintegrates.

Hewlett-Packard, Dell, IBM, Fujitsu and Cisco sold the most servers during the fourth quarter of 2012, according to data released by IDC. But it's the "other" category that's scaring the pants off the others. That's a throng of second-tier and "white box" server vendors who collectively sold 879,711 servers during the quarter. Number one HP, by contrast, sold 663,598.

Look no further than IBM's reported plans to sell off its low-end serve business to Lenovo — likely Big Blue's attempt to escape an increasingly commoditized market while it still can.

True, IBM, HP, and Dell still each pull in more revenue than the the $1.86 billion amassed by the "others" horde. But that's cold comfort, given that low-cost unbranded servers are quickly eating into their markets. The commoditization trend is getting a hard push from companies like Google, Facebook and Rackspace, who are busily designing and building their own servers to power their huge data centers.

Server manufacturers are trying to shield themselves with software, services and support, three defenses against the Mongol horde of white boxes. Arguably, though, that hasn't worked for Big Blue, at least at the low end of PC-based servers. And there's no particular reason to think it will save HP or Dell over the long term, either.

On the surface, none of this will really disadvantage consumers or developers. It really doesn't matter to end users whose name is on the boxes that power Netflix, Evernote, Apple's iCloud, or Amazon. And there will always be a need for some sort of server, from somebody. In fact, commodization is an indirect but very real plus for users, since it lowers costs for Web providers, making possible  an increasing lineup of innovative, and often free, Web services.

But it's definitely bad news for the server makers themselves. 

This Has Been A Long Time Coming

Traditional PC manufacturers like IBM, HP and Dell have taken one hit to their businesses after another over the past decade or so. Desktop PCs started declining in price with the rise of the Internet, reducing the need to upgrade PCs. Then laptops ate further into that market as they began to rival desktop power while offering mobility.

Finally, tablets and smartphones tapped the cloud for computing and location, and far more cheaply, limiting the need for people to buy expensive laptops.

That left PC-based servers — and that market is now under siege as well. The days of the mainframes came and went, and most server infrastructure now runs on the Xeon processor, Intel's PC processor optimized for the enterprise. Traditional mainframe processors — IBM's Power, Oracle's Sparc — retreated to the ivory towers of research computing. Meanwhile, companies like former Taiwan motherboard makers Supermicro or Asus realized that they can assemble a notebook or server just as well as a Dell or HP, and for less.

As prices of traditional PCs fell, hardware makers turned to new tactics, loading up new machines with "crapware" ranging from trial versions of AOL to antivirus programs to games. Consumers hated it, but the revenue crapware provided, directly or indirectly, helped keep hardware makers afloat.

Unfortunately, there's no way to duplicate that strategy in the server space. IT managers don't want servers cluttered up with Adobe Flash, Cyberlink PowerDVD, Roxio Creator, or any of the other bloatware that Dell places on its PCs. They do, however, want some help just making sense of it all.

Geek Squad On Steroids

Enter "solutions," the jargon that dominates enterprise discussions. Suddenly, the PC turned server makers weren't selling a PC, monitor and printer; they were packaging together a server, associated storage, a network switch, security, migration, and engineering services and support to tie it all together. Put extremely simplistically, an enterprise solution is everything that the Geek Squad offers, just scaled up by orders of magnitude.

In 2012, for instance, Dell bought Clerity in order to help Dell Services "reduce the cost of transitioning business-critical applications and data from legacy computing systems and onto more modern architectures, including the cloud." In the PC space, that's called dumping the contents of Mom and Dad's old PC onto a USB key and loading it into Google Drive.

Then again, Dell's enterprise "solutions" business climbed 4 percent last quarter, and pulled in $19.4 billion for its last fiscal year — about a third of its revenue. Cha-ching.

The central idea of the enterprise solution isn't the packaging. It's the customization, and the investment. Hewlett-Packard, for example, offers a fairly substantive list of industry-specific solutions for aerospace, automotive, and media, among others, with each pitching an additional value-added service. This was a tactic the same companies never really deployed in the PC market, perhaps because they never saw the need — or couldn't justify the investment — in designing PCs optimized for, say, tax professionals. 

Adding Value to a Commodity Business

Still, the "other" category is compelling evidence that a sizeable portion of the market seems to be unwilling to pay hardware makers for their services. Instead, they're pooling their resources. Massing and deploying large arrays of commodity hardware is the underlying principle behind everything from Hadoop — an open-source project for managing huge stores of data across distributed commodity servers — to the Open Compute Project. In that sense, the commodity server business is thriving.

And the PC makers themselves are helping it along, believing they can surf the trend by offering software and services on top of commodity hardware.

Dell's Data Center Solutions business, for instance, is a small but growing custom solutions business within Dell's larger server sales business. In January, Tracy Davis, the general manager of Dell DCS, attended the Open Compute Summit, whose principles include stripping the "vanity" logos from servers and replacing them with as much cost-optimized hardware as possible. Why would Dell participate in a forum seemingly designed to kill it off?

Davis told me that Dell is able to engage — and sell — everything from engineering services to Dell's ability to buy components all over the world. That's a competitive offering, not necessarily reflected in the bottom line, that still adds value to Dell's business versus a no-name, commodity server maker.

The Writing Is On The Wall

In some sense, things came to a head this week after CRN and other outlets reported that IBM was in talks with Lenovo to sell its low-end X86 server business for between $5 billion and $6 billion — an eerie parallel to the way Big Blue sold off its ThinkPad notebook business to Lenovo years ago. IBM hasn't confirmed or denied the talks.

"Assuming IBM divests [its] low end (low margin) x86 biz to Lenovo, HP and Dell should be concerned because Lenovo can make [money] and disrupt [the] space,” Matt Eastwood, an IDC server analyst, wrote on Twitter. The idea, Eastwood and others suggested, was that IBM couldn't squeeze money from a commodity business. Lenovo can.

But what's the commodity? Generic servers? Not necessarily. 

Solutions Can Be Commodities, Too

IBM's highest-profile service is Watson, the natural-language technology that beat several past "Jeopardy" champions and is being used in financial services and to help treat cancer patients. Watson and other related services run on servers based on its Power chips, not x86. Yes, IBM deploys a whole slew of services on its line of x86 servers — but they're awfully similar to what everyone else does, too.

Eventually, companies like ARM say, we'll all be running servers on the sort of low-power processors that power our cell phones, with the Web's basic functions — serving up static Web pages, for example — running on cheap, purpose-built machines. These aren't just commodity servers; these are commodity solutions. Meanwhile, companies like Google and Facebook are quietly building their own custom servers to fit their own, specialized needs.

Here's what IBM may be thinking. Since its highest-value, unique service offerings run on Power, they justify further investment. And if IBM can offer uniquely high-value services on top of the cheaper x86 boxes, then it should hold onto those, too. But if the company can't see customers turning to IBM for those solutions — either now or in the future — then IBM's justified for ditching them while the getting is good.

IBM led the way in pulling away from the PC in 2004, a controversial move at the time that now seems more than justified. If IBM takes the same approach with its x86 server business, it may be a similar harbinger of doom for other makers of x86 servers.

Image courtesy of Hewlett-Packard

IBM has no stomach for low-margin businesses, which is why Big Blue may be ready to dump its commodity server business — i.e., servers that run on Intel-compatible "x86" processors. If the reported talks with Lenovo lead to a sale, the move would mark IBM's final break with the low-end computer business.

A Win-Win

The deal would be a win-win for both companies. Lenovo, which bought IBM's PC business in 2005 for $1.75 billion, would immediately become the third largest maker  of x86 servers, behind market leader Hewlett-Packard and runner-up Dell. Thanks to its market clout in its homeland, the Chinese company has risen to become the second largest PC maker worldwide, according to the latest numbers from IDC. 

Adding x86 servers to its portfolio makes perfect sense for Lenovo, which has shown in PCs that it can do well in a low-margin, commodity market. For IBM, the opposite is true. The company's strength in hardware is in selling expensive — and profitable — mainframes.

IBM's mainframe business is the reason the company leads the global server market, at least in revenue terms. To give you some sense of how expensive these systems are, IBM's "System z" mainframe represented more than 12% of all server revenue worldwide in the fourth quarter. Because of a refresh in the product line, along with the introduction of new products, such as the zEnterprise, revenue from IBM's mainframe business rose almost 56% year over year in the quarter, according to IDC.

"Although revenue results for System z are traditionally heavier in the fourth quarter, this accelerated acquisition shows the breadth and depth of the IBM mainframe installed base," Jean Bozman, analyst for IDC said in a statement.

Lenovo would be a good buyer for IBM, because it doesn't compete in any of the markets IBM cares about, namely software and IT services. That wouldn't be the case if HP or Oracle were the buyer.

Disruption In Server Market

IBM may also have decided it wants no part of the disruption heading for the server market like a freight train. The increasing number of companies adopting cloud computing will mean fewer server sales, Larry Dignan points out at ZDNet. In addition, Internet companies with large server farms, such as Facebook and Google, buy customized white-box servers, which can't be good in the long term for traditional sellers, like HP, Dell and IBM.

While no one outside of IBM or Lenovo know how much the business would fetch, someone familiar with the talks told Bloomberg that the price would range from $2.5 billion to $4.5 billion, depending on the assets and liabilities included.

Lenovo Is Fired Up And Ready To Go

Not everyone agrees that IBM would be doing itself a favor by selling its x86 business. Gartner analyst Sergis Mushell says that without x86, IBM only non-mainframe servers would be its lineup of machines that run its Power processors — and that demand for those products is shrinking.

In other words, IBM would miss out on the opportunities to build systems based on x86 "while [its Power] architecture's ecosystem is shrinking," Mushell said. "Do you see how it would not make a lot of sense?"

Lenovo, meanwhile, is hungry to move beyond the PC market. The company announced last year a partnership with EMC in which Lenovo planned to introduce x86 servers that would include EMC storage systems. As part of the deal, Lenovo agreed to sell EMC networked storage products in China.

Given the jumpstart it would get from owning IBM's x86 business, Lenovo may be willing to make an offer that's hard for IBM to refuse.

Image courtesy of Shutterstock

As Hewlett-Packard, Dell, Lenovo, and other manufacturers nervously place bets on the PC, server, and tablet markets, they're playing more and more with Intel's chips. And that means one thing: Intel stands to win no matter what.

Intel yesterday reported a 25% profit decline on revenue of $12.6 billion, which the company blamed on the general blahs plaguing the PC market. But two numbers stood out: a 6.6% drop in PC microprocessor revenue, and a 7.5%  increase in revenue from data centers.

Playing Both Ends — Maybe All Three

What does this mean? At the moment, PC sales are in free fall as consumers rush toward tablets. But as customers snap up mobile devices, tapping into cloud-hosted apps, demand for the servers that power those data centers increases.

Meanwhile, Intel is moving farther into phones, tablets, and networking, where further profits beckon. The bottom line is this: if consumers chase mobile apps, Intel will be there, powering cloud data centers. If they stick to the PC, or shift to new lightweight ultrabooks, Intel stands to benefit thanks to its 80+ percent market share. And if Intel can convince more phone and tablet makers to buy into its chip offerings, it'll win there, too — though that's much more of a gamble at the moment.

Intel CFO Stacy Smith told analysts that the company's second-half outlook looks stronger than expected because of two things: a stronger macroeconomic environment, which would boost overall spending, plus Intel's presence in PCs, servers, phones and tablets. 

To that, Intel chief executive Paul Otellini added a third component: price. "We have a certain spec for ultrabooks, and that is the product that Stacy said is going to be centered at as low as $599 with some [products] to $499," Otellini said. "If you look at touch-enabled Intel based notebooks that are ultrathin and light using non-Core processors, those prices are going to be down to as low as $200 probably."

Intel's Hole Card: Mastery Of Moore's Law

Otellini, who will retire in May, can fairly be criticized for not investing in tablets and other mobile devices earlier. But from an operational standpoint, Intel is winning. The company continues to leverage its core asset: manufacturing, creating a ripple effect that continues to carry the company into new markets.

In May, Intel will launch "Haswell," its next-generation 22-nm chip. Rival AMD is a generation behind, at 32 nm. This forces AMD to out-engineer Intel — again — in terms of chip design to keep up, and AMD arguably hasn't done that. "This leadership in materials science and manufacturing technology is the foundation on which our future success will be based, arming us with the world’s lowest power and lowest cost transistors," Otellini said, and he's right.

Normally, a drop in PC consumption, and thus lower manufacturing demand, would imply a decline in revenues. Not so. Smith said that Intel simply pulled older manufacturing equipment and accelerated a shift toward its next milestone, 14-nm manufacturing, and saved $1 billion in capital costs in the process. It also struck a "foundry" deal with Altera, in which it agreed to manufacture Altera chips on unused Intel equipment. If demand picks up, Intel can simply turn on production lines again.

Mobile and Tablets Still Hold Potential

After fumbling its StrongARM technology in 1997 — the processor architecture which now powers basically every phone on the planet — Intel shocked many by announcing X86 phone designs with Lenovo and other Asian manufacturers in 2012. Is Intel poised to take over the phone market? Absolutely not. But simply demonstrating the capability makes it a company to watch, and its reach may slowly grow over time.

So far, the "Clover Trail" Atom chip Intel debuted last fall for a new generation of convertible Windows tablets has barely left a ripple, hampered as it was by poor computing performance and a general disdain for Windows 8. In phones, however, Intel is combining Clover Trail with an applications processor and an LTE baseband chip into what's known as a system-on-a-chip, a tidy all-in-one package. First-quarter tablet volume doubled, and Intel expects it to double again — from a little to a little more than a little, one might expect. Still, it's a start.

Tablets, though, could be Intel's future. In the second half of the year, Intel plans to launch "Bay Trail," a quad-core Atom chip. (Intel brands its PC processors using the "Core" name; the non-Core chip Otellini referred to inside the $200 PCs is almost certainly Bay Trail.) If that's true, a $200 Windows-based tablet almost moves into impulse-buy territory. 

What Intel's irresistible progress in manufacturing technology means is that it almost doesn't matter whether Clover Trail or Bay Trail are successful. Intel should enable the combination of performance, power, and/or price that should offer Windows tablets some true competition to Android and iOS. Will ARM be able to keep ahead? If we're talking $200 price points, how much will it matter?

Meanwhile, Intel's hold on the enterprise market remains secure, as the vast majority of servers are powered by Intel X86 chips. Here, too, ARM chips have declared war, claiming that their low power offers a more cost-effective solution to Intel's power-hungry Xeon chips. Intel has deployed "Centeron," an optimized Atom chip for servers, in response. And software-defined networking, which steals some of the intelligence from a network router or switch and puts it inside a server, benefits Intel, too.

What Intel does best, though, is double down and double again, using manufacturing to make up for any shortfalls in design that its competition might otherwise exploit. It might not be the most elegant solution, but so far it's proving brutally effective.

Image source: Intel

Magicians work their performances with a combination of talent, slight-of-hand and a lot of misdirection. On stage, a well-orchestrated misdirect is enough to pull off the best illusions. In the corporate world, it can sometimes work well, too, as Oracle seems to have discovered with their school-yard baiting of IBM - a move that is distracting some away from Oracle's missed third-quarter earnings.

To see the misdirect in action, let's briefly look at the timeline of events:

• Mar. 20: On Oracle's third-quarter call, the company reports a significant earnings miss, which initially drove the company's stock down 8% in trading. Executives on the call noted a 2% drop in new software sales and Internet-based software subscriptions in the quarter, a problem they attributed to a rapidly expanding salesforce (i.e., blame it on the new sales folk).

(See also Oracle's Big Miss: The End Of An Enterprise Era?)

• Mar. 26: Oracle announces a new SPARC T5 line of mid-range servers (along with some mainframe M5 boxes), which starts yet-another line of questioning about the viability of using expensive multi-threaded servers in an age where Intel dominates single-threaded commodity servers and ARM-based servers are on the horizon. It's a familiar question, since some were asking it when Oracle pushed out its T4 line in 2011. But such questions are soon forgotten, because…

• Still Mar. 26: During the SPARC announcement, Oracle CEO Larry Ellison, well known for his genteel manners and discretion, throws down some smack talk on IBM's more expensive server line, with the stats to back up their claims, claims such as "[t]he SPARC T5-8 server delivers 1.4 times better performance than the next best single system result, the 32-processor IBM Power 595, at one-fifth the price/performance." Wait for it… wait…

• Mar. 27: IBM Marketing Manager Steve Sibley chomps at the bait, firing off a response to Business Insider that Oracle's comparisons are wildly off the mark, since they are comparing brand-new Oracle boxes to IBM setups that are completely different configurations and also three to five years old. Sibley also reminded BI that this isn't the first time IBM's had Oracle pull down wild claims in advertising and public statements about the performance of its servers.

(Check out Oracle Has Problems Telling The Truth In Its Advertising)

• Mar. 28: Oracle doesn't back down, as John Fowler, Oracle's executive vice president of systems, emails Business Insider: "IBM customers are being wildly over-charged for the performance they're getting."

Note, please, that in all of this hullaballoo, no one is talking much about Oracle's missed earnings, and any conversations about Oracle's hardware are in the context of comparing them to IBM's older hardware products - right where Oracle wants those conversations to be.

"A la peanut butter sandwiches!", as The Amazing Mumford would say.

If this were a one-time thing, it would be merely something of note. But a pattern seems to be forming with Oracle and their product announcements: it's not enough to just hold up their own hardware and software features, but they habitually have to take pot-shots at anyone else in the space as well. That's all part of doing business, of course, but Oracle appears to just make things up or cherry pick data to create a scenario that makes them look the best.

Sooner or later, though, such tactics may wear thin on potential customers and even existing customers. If you treat marketing as one big magic trick, how long before people start to wonder if what you have to offer is real, or an illusion?

On Tuesday, Dell faced Wall Street analysts for what could be the last time, as Michael Dell and a collection of investors prepare to take the company private. And though Dell Inc. reportedly exceeded Wall Street's expectations, the results were disappointing overall. And in some ways, that's a good thing.

Dell revenues fell 11% to $14.3 billion. Profits were down, too: 31% to $534 million. Dell's consumer business fell by a whopping 24% to $2.8 billion; The slogan "Dude, you're getting a Dell" is now a distant memory.

The earnings call is a unique event in American business; although chief executives occasionally deign to hear questions from business reporters, rarely do they sit down with their upper management and submit to questions about their past and future financial and operational performance. Calls following quarters in which a business dramatically exceeds expectations, taking a drink each time an analyst congratulates execs ("Great quarter, guys!") will usually result in a long nap under one's desk.

After a lousy quarter, on the other hand, analysts feel unusually liberated to ask the pointed questions that should always be asked. And - hurray! - some of them did their jobs on Tuesday.

Here are the five questions that Dell will need to answer going forward:

1. Does Dell Belong In The PC Market?

In many ways, this is the same question that Hewlett-Packard faced in the wake of Leo Apotheker's decision to shop the PC business, Meg Whitman's decision to retain it, and then the ultimate reorganization into a combined printer/PC business. With very little room for adding value besides bundling a printer with a PC, adding the crapware software that users hate, and striking out into relatively untested waters such as ultraportables and tablets, the answer seems to be: "Yes, but barely."

Dell's strengths are its XPS tier and its Latitude line of business PCs. But as Dell chief financial officer Brian Gladden noted, the growth continues to be in tablets and in the low-value desktop and notebook space - both areas where Dell, seeking higher profits, has consciously avoided.

2. Is There An Opportunity To Refresh Older Corporate PCs?

Yes, definitely. And that's the primary reason Dell won't bail out of the PC market any time soon - it has established longstanding ties with corporate America. This was one of the most telling quotes of the call:

"I think that’s really tough to get at, but the data that we’ve seen would suggest there’s still somewhere in the range of 40% of the corporate installed base for PCs that is XP or Vista that needs to be upgraded," Gladden said. "So that’s, I think, pretty consistent with the data that we see for our installed base, and for what we hear from our corporate customers."

Corporate IT departments rarely, if ever, update a PC operating system without refreshing the hardware, too. Microsoft may have to worry about corporate customers turning to Windows 7 rather than 8, but Dell doesn't care either way. And with support ending for Windows XP in April 2014, Dell knows there's a windfall ahead.

"All the data that we’ve seen, all the conversations we’ve had with customers, would lead us to believe that there’s still a significant refresh activity that has to happen in the next 12-14 months," Gladden said.

3. Is BYOD (Bring Your Own Device) The Answer?

Yes. Or it was in December, when Michael Dell said exactly that. 

"And in the customer conversations that we’ve been having, the interest in Windows 8 is quite high, even with commercial customers, who would normally wait a few releases to adopt the new versions," Dell said. "What we’re seeing here is really an immediate need, because CIOs are worried about the ramifications of a BYOD world. With Windows 8 products... we’re pleased with the incredible experience that they expect, while you get the security and versatility and reliability that your enterprise really requires."

Since then, Microsoft has reported terrific Windows 8 numbers, although there has been some suspicion that the company hid behind previouslysold Windows 8 licenses. Still, Dell and the rest of the PC industry will certainly help try and make Windows 8 a success. Ultimately, however, Michael Dell has bet his farm on Windows, while other hardware makers, like Samsung, have diversified into Android and phones. Time will tell who made the right choice.

4.) Will Server Customers Keep Buying From Dell, Or Roll Their Own?

If you're not following the datacenter market that runs the cloud services we use and love, you're may be unaware that Facebook has pioneered an industry-wide program called Open Compute, which publishes detailed specifications on building your own servers via components from no-name manufacturers. Facebook recently said that a major European datacenter would be constructed entirely from these "white box" servers, and companies like Rackspace have also signed on.

Revenue-wise, Dell’s Server and Networking Business unit grew 11% in the last quarter, to $9.3 billion. But ISI Group's Brian Marshall asked one of the key questions: given the Open Compute model, will the trend continue? Gladden waved off the question.

"It is relatively isolated to a few number of large-scale customers who can make the economics work, and given that, we’re still seeing strong growth in that business, and significant opportunities to continue to grow the hyperscale business," Gladden said. "So I don’t think it’s a new dynamic."

Gladden's right; most companies are not going to exert the time and effort to design their own servers, Open Compute or no. But over time, this may eat into the server businesses of Dell and others.

5.) What Effects Will Virtualization, Consolidation and the Cloud Have On Servers?

Virtualization, where a number of "virtual servers" can share computing resources, helps effectively consume underutilized servers, especially older hardware. Some of the older hardware can be retired, as data centers "consolidate" their hardware and run virtual machines on top of them to maximize their use. At the same time, as more cloud services are deployed, the number of servers they require goes up. Unfortunately, analysts have reported that the number of servers sold has flattened - the trends of consolidation and virtualization are holding down sales, and revenues are actually decreasing.

For a long time, notebooks became the escape route out of the quicksand of commoditization that has dragged the industry down. Then servers were the answer. Now, they're apparently sinking into the mud, too.

Dell may in fact continue to provide updates to Wall Street as it goes private; it hinted as much when it talked about a fiscal first-quarter earnings release. But as the company moves into the financial shadows, away from public scrutiny, it did not provide any guidance for the future. One can wonder whether its outlook is equally cloudy.

Web developers using OS X, take note: If you want to create a fully contained server stack in which to build and test your latest ideas on Joomla, Drupal or WordPress, you're just one click away from creating such an environment.

Four Different Choices

Within the App Store for OS X can now be found four such stacks, courtesy of BitRock's Bitnami, a free software service that enables you to install various software stacks either natively on Windows, OS X or Linux; as a virtual machine in VMware or as an Amazon Cloud instance.

Full disclosure: I've been a Bitnami fan for a long time. The stacks it offers, which include Alfresco, ownCloud and SugarCRM, are very easy to install and are perfect for fast setup when I want to review software or slap a together a website.

The stacks offered in the App Store include Joomla, WordPress, Drupal and a generic MAMP stack (Mac, Apache, MySQL, PHP) - all popular website platforms that are installed natively on your Mac machine.

Bitnami stacks installed natively are not installed as they would be if you built the software in the stacks by scratch. Instead of code getting installed all over the place, the binaries for the stack are all placed inside one directory, completely self-contained.

A Different Kind Of Walled Garden

The Bitnami stack from the App Store, it seems, are walled off even more, according to reports from users. The App Store's sandboxing apparently makes configuring the software a little harder than it normally would be, so if you're going to do extensive development with these stacks, users are recommending you visit Bitnami and get the native installation packages from the company directly.

This is not to decry the App Store's Bitnami stacks. I pulled down the Joomla stack, installed and configured it, and was ready to work with it in minutes. I can play around with themes and extensions in Joomla to my heart's content. Best of all, there was little to no resource dragging on my system, which I sometimes experience when I run one of these stacks as a virtual machine in VMware or Parallels.

Serious developers may indeed want to pull down the images straight from Bitnami, or better yet, install one of these stacks as a full-on Amazon Machine Instance on the EC2 platform and create an eventual production version of the stack you're creating.

It's too easy not to.

Lead image courtesy of Shutterstock.

Remember SUSE? Way back when it was the cutting-edge Linux distribution, and held its own with Red Hat. But that was a long time ago, long before Microsoft adopted it as its pet and Attachmate took it over as part of its Novell acquisition. With Red Hat dominating the enterprise Linux server market, Canonical owning the Linux desktop market, and Google's Android running roughshod over everyone in the mobile market, what, exactly, is left for SUSE?

In The Clouds

Cloud, perhaps? After all, Alan Clark, director of Industry Initiatives, Emerging Standards and Open Source at SUSE, and a friend of mine, was elected in 2012 to chair the OpenStack Foundation board. OpenStack seems to have real momentum, but ever since Red Hat got involved, it's hard to see OpenStack turning out much different from Linux, where Red Hat wins in part because it's such an active contributor. Already Red Hat has gone from a somewhat light contributor to the third-highest contributor after Rackspace (OpenStack's founder) and HP.

SUSE? It barely makes the list of top-10 contributors.

In open source, being the source of the code is more important than owning the source code, and Red Hat is on pace to be the dominant contributor to OpenStack. This can't be comforting to SUSE. 

Even if we look at the Linux distributions that individuals run on Amazon Web Services (AWS), Rackspace and other public clouds, SUSE shows up as a rounding error, with Canonical's Ubuntu commanding the market and Red Hat Enterprise Linux clone CentOS coming in second. It's telling that when HP, a longtime SUSE supporter, had to choose an operating system to power its own public cloud, it chose Ubuntu. 

All of which leaves SUSE in a precarious position. 

On Life Support?

No, SUSE is not dead yet.  As longtime Linux pundit Steven J. Vaughan-Nichols told me, 

@mjasay From the outside looking in, SUSE still has some presence on servers, and it still seems strong on mainframes.

— sjvn (@sjvn) January 25, 2013

This is true, but in a conversation with a former SUSE employee who is familiar with SUSE's past and current performance, revenue from SUSE's hardware partners like HP and IBM has been constant but stagnant over the past few years. As he puts it, these longtime SUSE partners want a hedge against Red Hat, but they know that their businesses largely depend upon Red Hat. So they give SUSE just enough business to keep it alive.

This doesn't tell the whole story, though. Other sources inside the company tell me that last year SUSE exceeded its sales targets. Last year, as Forrester notes, SUSE brought in $200 million in revenue and expects to ratchet that up to $234 million.  Life is easier for SUSE now that it has jettisoned the need to upsell Novell's (pretty tired) management products. 

Post-Novell, all SUSE needs to worry about is Linux, and SUSE Linux has always had a reputation for serious quality. Now that it has seriously focused itself on the enterprise server market, eschewing erstwhile "sexy" markets like mobile, it's a much more coherent story to tell would-be customers. It remains relatively strong in Europe, as Paulo Frazao highlights, and its role as a hedge against Red Hat puts it in a good position with VMware, in particular, as Ian Waring suggests.

But even as a Red Hat hedge it plays second fiddle to CentOS, of which I was reminded by Kevin Schroeder. No, not with the server vendors, who generally avoid CentOS in an attempt to placate Red Hat. But over the past few years I've seen very large enterprises shift applications, including mission-critical applications, to CentOS as a way to cut costs. And in terms of general interest in the two platforms, well, a chart says a thousand words:

Still, this doesn't help SUSE. While I believe the company is wise to focus on the enterprise market, and not distract itself by chasing mobile or even desktop markets, doing so leaves it to compete against its old nemesis, Red Hat, without any compelling reason for the market to drop its preferred RHEL distribution for the Avis of Linux.  SUSE may well "try harder," but all it seems to have earned for its troubles is a permanent position as a distant, second-place hedge against Red Hat, whose lead continues to grow in the markets about which SUSE cares most. That's a dangerous position for any company.

Lead image courtesy of hotamr. 

At the heart of a PC or server, the microprocessor beats thousands of times a second. And until now, if you wanted to exchange it for a newer, more powerful model, the transplant was almost doomed to fail. Until Wednesday.

Yes, PC owners can upgrade an AMD processor, for example, to another processor that uses the same electrical socket. But on Wednesday, Facebook announced "Group Hug," an awkwardly-named daughtercard that does something quite wonderful: allow a customer to replace an AMD chip with an Intel chip on a server- or even something completely different, such as an ARM processor.

For someone used to buying an off-the-shelf PC and replacing it wholesale a year or two later, this might not seem like a big deal. But DIY computer builders know that when it's time to upgrade, they find themselves holding onto components like hard drives, just because they're loathe to throw them away. But the combination of a PC motherboard and CPU often costs a couple of hundred dollars or so; in the server space, it can cost much more. The ability to "right size" a server with a CPU that's optimized for a particular task often requires selecting a server out of a catalog. And in a couple of years' time, it means selecting another server, rather than just upgrading a chip.

“I’m going to be able to take the older [components] out, and leave the ones still of value, still of use, in the rack," said Frank Frankovsky, director of hardware design and supply chain operations at Facebook, at the Open Compute Summit on Wednesday.

Open Compute: Enterprise Hardware Hackers

On a technical level, the Open Compute Project, as its known, is essentially a homebrewed server, from a custom chassis design to influencing the design of many of its components. From a business standpoint, the OCP challenged the traditional hardware hegemony. Just as homebrew computer users construct their own PCs, two years ago Facebook had the bright idea of designing their own servers. (Google has done this as well.) Facebook decided, however, to share what it learned with other hardware providers, improving the quality of the industry as a whole.

While hard drives use a serial ATA connection and graphics cards use PCI Express, each CPU provider has used their own socket connector and bus interface, preventing one microprocessor from being swapped for another. Group Hug places the microprocessor on a daughtercard or add-on board, communicating with the rest of the server via a generic x8 PCI Express interface. Exchanging an AMD or Intel chip would be impressive enough, but to be able to add in an ARM chip - which runs an entirely different instruction set - takes this to a new level. 

Even more impressively, AMD and Intel have signed on, as have two ARM providers, Applied Micro and Calxeda. One would think that AMD, for example, would try and build in roadblocks to being swapped out for an Intel chip. "But each manufacturer is so convinced that they're the best that I'm not sure this matters to them," said Nathan Brookwood, an analyst with Insight 64. 

It's not clear, at the moment, whether or not Facebook and the OCP will have to develop protocols to allow the processor's instructions to essentially be "abstracted," or packaged in a generic fashion that could be understood by the other components within the server. Like the "Graph Search" technology Facebook showed off a day earlier, Frankovsky implied that there was more work that needed doing before the "Group Hug" boards could be considered a reality. Still, manufacturers showed off Group Hug hardware here at the show, indicating that the future isn't that far off - assuming, of course, that both AMD and Intel choose to offer more than just a small number of Group Hug parts. 

One might argue that the declining PC market has forced microprocessor designers to accept solutions which may not have been palatable a few years ago. By designing CPUs with a generic interface, Brookwood pointed out, it might mean that customers would be far more willing to swap new ones out, increasing both CPU turnover and sales. And, at least in the server market, traditional hardware makers like Dell and HP have adopted the role of solutions provider, selling value-added, high-margin software services in conjunction with servers that are always in danger of commoditization.

Is This The Future Of The PC?

It's not hard to see this trend making its way to the PC, where finally enthusiasts will be able to swap out which parts they want. The question is, of course, whether there will be a viable enthusiast PC market by that time. 

Still, Group Hug represents a revolution of sorts. A decade ago, consumers and enterprise customers alike bought a collection of parts that AMD, Intel, and its partners handed down. As a result, there was little difference from one PC to the next. With Group Hug, the pendulum has swung back to the customer, who will be able to chop any product that doesn't cut it, and replace it with one that does. 

While Hewlett-Packard says it "continues to evaluate" the sale of underperforming businesses, the company's cash flow problem will make the shedding of assets unavoidable. So what's likely to head to the auction block? Everything from notebooks and desktop PCs to Itanium servers and tape drives that have been draining assets could be on the market.

A Breakup Alternative

For Chief Executive Meg Whitman, selling off pieces of the crippled tech giant would be a much better alternative to breaking up the company. Whitman has opposed the latter option, starting with her decision in 2011 to nix a proposal by her predecessor, Leo Apotheker, to spin off the company's personal computer unit.

Since then, Whitman has ignored Wall Street analysts who say shareholders would be better off if the company spun off the division that sells PCs and printers from the one that sells software, hardware and services to companies.

As a less dramatic alternative, getting rid of businesses draining the company's limited resources, would help HP make better use of limited cash. In fiscal 2012, HP's free cash flow dropped to $6.9 billion from $8.1 billion the previous fiscal year, according to The Wall Street Journal. That's a trend that could spell trouble if not stopped. Without cash, a company will find it difficult to develop new products, make acquisitions, pay dividends and reduce debt.

Getting rid of underperforming businesses is one way to improve cash flow and avoid splitting the company. "Everybody zeroes in on printers and PCs as the things they should potentially sell, and quite frankly, there's not really a logical buyer for either of those businesses," Crawford Del Prete, analyst for International Data Corp., said. "And, those businesses generate a significant amount of cash, which Hewlett-Packard needs right now."

HP-UX And More Must Go

A more logical sale would be the Itanium server business. HP has spent a lot of money trying to drive sales of its HP-UX Unix server that runs on that chip architecture, while the business continues to shrink. In 2010, Microsoft said it would drop support for Itanium and Oracle said a year later it wants to do the same.

Another candidate for jettisoning is HP's low-end IT outsourcing business, which was included in the 2008 acquisition of Electronic Data Systems. Earnings from the services business has been falling, and last August, HP said it would write off $8 billion in goodwill from the EDS purchase.

Last year, General Motors, a major HP customer, said it would move away from outsourcing IT and take some work in-house. The announcement made industry observers wonder whether HP can handle those large-scale deals, Del Prete said.

Within HP's Personal Systems Group that makes PCs, workstations, tablets and printers, the company could sell the low-performing notebook and desktop PC businesses, which have been trumped in the market by tablets.

The low-end printer business that primarily serves consumers and small businesses could also be sold. However, printers are still used in emerging markets, so HP would be just as likely to hold off to see how profitable those markets become. "HP has a plan to drive those businesses, so I'd be surprised to see them get out," Del Prete said.

Finally, tape drives used for long-term data storage is a candidate within the company's enterprise servers, storage and networking division. Such a low-margin business would be best left to IBM and others with larger stakes in the market.

HP likely has other losers within its product lines that it would be better off without. Whitman should act quickly to get rid of the chaff and focus resources only on the profit generators.

Image courtesy of Shutterstock.