In-Depth
The Enterprise Year That Was
From powerful mainframes to small smartphones, it was a year filled with change and challenges, contradiction and contrast for enterprise IT.
For IT professionals, 2010 was a year of contradiction and contrast. The mainframe market experienced a setback yet welcomed the powerful new zEnterprise system. IT focused on getting the best performance from big mainframes yet faced as never before the challenges of managing small mobile devices. More data center components were virtualized even as more processing and storage migrated to the clouds.
Let’s take a look at some of the highlights of the year for enterprise IT.
Mainframe Market Setback
Big Iron fortunes took a big step back in 2009, thanks chiefly to the parlous state of the economy. Demand for servers was down across the board, of course, but even after sales of commodity hardware had recovered, to an extent, by December of last year, demand for Big Iron hardware was still depressed.
This lackluster trend carried over into 2010, with mainframe demand depressed through both the first and second quarters of this year. IBM Corp. and others in the industry attributed this to shops holding out for Big Blue’s next-generation mainframe kit, which was supposed to ship by late 2010.
With news of a major mainframe refresh in the offing, some suggested, shops were postponing new mainframe purchases in anticipation of a bigger -- and altogether MIPS-ier -- zNext deliverable from IBM.
Big Blue’s mainframe market setback significantly affected both its bottom line and its stature. Sales of pricey System z mainframes had longed helped to buoy IBM's share of global server revenues. With both mainframe and System p selling poorly, rival Hewlett-Packard Co. (HP) was able to catapult to the top of the server market heap: it grabbed the lead in Q1 and held it for the rest of 2010.
That's one reason IBM moved up its zEnterprise launch date -- slated for August or September -- to July.
Which begs an obvious question: was zEnterprise worth the wait?
Birth of a Behemoth
In a word: yes. The zEnterprise system that IBM announced in July really and truly isn’t like any mainframe that came before it. It packs 96 processing units (running at 5.2 GHz) into a single system, with a maximum of 80 processing units available for either Linux or z/OS workloads. That makes it about 60 percent faster than its predecessor, System z10. Each zEnterprise system can process more than 50 billion instructions per second -- while consuming about as much electricity as an Enterprise Class (EC) System z10 mainframe.
zEnterprise isn’t a System z-only proposition, either. Big Blue announced new offerings -- its zEnterprise BladeCenter Extension and zEnterprise Unified Resource Manager -- which permit mainframe, POWER7, and System x workloads to share resources, all managed -- by zEnterprise, of course -- as a virtualized system. For the first time, shops can use zEnterprise to manage a “cluster” of tens of thousands of virtualized servers as a single system.
IBM’s zNext refresh appears to have done the trick. According to Gartner’s Q3 server market study, for example, sales of systems powered by silicon other than x64, Itanium, or Unix-oriented RISC architectures were up by just over 8 percent. That tally does include revenues from non-System z platforms: in addition to IBM’s System i platform, Unisys Corp. and several other vendors market proprietary mainframe CMOSes, but System z accounts for a big chunk of it.
IBM and other mainframe boosters expect Big Iron to recover nicely in the coming year. Mainframe market surveys from both BMC Software Corp. and CA Technologies Inc. show Big Iron at least holding its own -- with at least half of mainframe customers planning to add capacity over the coming months.
The PC Problem
Back in August, Intel Corp. ponied up almost $8 billion for security software specialist McAfee Inc. In one sense, you could interpret Intel’s gambit as a kind of doubling-down on the Wintel client/server paradigm that -- for three decades -- had reliably fueled its expansion. McAfee, after all, has a thriving anti-virus business: along with Symantec Corp.’s Norton-branded products, its brand is basically synonymous with anti-virus, particularly in the desktop space.
Intel’s acquisition of McAfee took place in the Year of the iPad, and the iPad, most agree, is a category-redefining deliverable. It’s also -- potentially -- a paradigm-busting proposition.
Announced in April of this year, the iPad was an instant sensation: by September, PC OEMs such as Hewlett-Packard Co. (HP) and Dell Computer Corp. -- to say nothing of mobile giants such as Research in Motion (RIM) Inc. and Samsung -- had announced iPad-like computing tablets of their own.
The iPad helped put the PC Problem squarely into focus. Tablet PCs are nothing new, of course, but the iPad isn’t really a Tablet PC: it eschews that platform’s bulky footprint and thick-client operating system in favor of a sleek physique and what might be called a “svelte-client” operating environment.
Hallelujah, say many IT professionals. IT had to suffer the shortcomings and inadequacies of PCs precisely because users wanted PCs. They didn’t want dumb terminals, first- or second-generation thin clients, Java stations, network computers, and the like. They wanted both Excel and Minesweeper; a Web-based front-end to an old CICS application and People.com. Back in the late-90’s, when Messrs. Ellison and McNealy were trumpeting the inevitability of the network computer, the client/server Powers-That-Be had a simple response: users won’t stand for it.
That used to be the case, but at some point -- probably at some point this year -- it ceased to be so. Maybe it took some combination of the iPad, the increasing ubiquity of smartphone devices (particularly Apple Corp.’s iPhone and Google Inc.’s Android), the success of other non-traditional devices (such as Amazon.com’s Kindle or Sony’s Reader), the established emergence of desktop virtualization technologies, and the mainstreaming of cloud computing.
Regardless, the “personal computer” suddenly started to look cumbersome, bloated, wrongheaded, and inconvenient
Which brings us back to Intel and McAfee. Although McAfee made its name in the combined desktop security and antivirus space, it has since diversified its portfolio. Like other players, McAfee broadly focused on “endpoint security,” a category that involves non-traditional -- or non-PC -- devices, including smartphones, embedded or thin-display devices, and virtually any connectible asset of any kind. Earlier this year, for example, McAfee acquired mobile security specialists Ten Cube and Trust Digital, broadening its endpoint security portfolio.
For this and other reasons, industry watchers saw Intel’s acquisition of McAfee as less of a double-down on an established market and more of an attempt to position itself in an emerging market.
Of course, $8 billion isn’t chump change, even for so large a company as Intel. The desktop computing times are a-changing. Smartphones are increasingly ubiquitous -- by 2012, IDC reported in December, combined sales of smartphones and tablet computers will surpass those of the PC.
Meanwhile, HP, IBM, Cisco Systems Inc., EMC Corp., VMware Inc., and other players stepped up their efforts on the virtual desktop front. In November, for example, Cisco -- which entered the computing hardware segment two years ago, with its Unified Computing System (UCS) pitch -- partnered with EMC and VMware to deliver a turnkey desktop virtualization offering.
X64: Meet the New Boss
Also underscored in 2010 were the contrasting fates of the x64-based server -- long the upstart David to the RISC-based server’s Goliath -- and Unix.
In March, Intel Corp. released Nehalem EX, the latest revision of its Xeon silicon. With Nehalem, Intel was explicitly dreaming big: for the first time, mostly-off-the-shelf x86 systems would -- in some cases without the use of clustering -- scale to support to RISC-Unix-like configurations.
For example, at least two OEMs -- Bull Information Systems and IBM Corp. -- introduced Nehalem EX-based systems with 1 TB or more of main memory. Big Blue, in fact, took the memory configuration crown: it introduced a server -- the x3850 X5 (or “MAX5”) -- that could support a whopping 3 TB of memory.
X86 -- or x64, as it now stands -- took server market bragging rights from Unix a long time ago, but something else happened in 2010 -- or, more properly, in 2009. In the fourth quarter of 2009, for the first time ever, sales of x86 servers accounted for more than half of all server revenues.
This shift -- a sea change -- took place in 2009 but wasn’t recorded until 2010, when market watchers Gartner Inc. and IDC published their Q4 reports.
Meanwhile, the fortunes of RISC-Unix continued to decline: it was down in Q1, sunk further in Q2, and was just-as-down in Q3.
How down? Roughly speaking, from quarter to quarter, unit shipments were off by double-digits and revenues by close to double-digits at least.
Unix -- or the non-Intel-based market, which consists (by and large) of Unix systems powered by RISC or Itanium processors (as well as of Big Iron platforms powered by specialty mainframe CMOSes) -- suffered an historic decline this year, too. For the first time in the 14 years that market watcher IDC has been tracking server market buying patterns, non-Intel-based platforms failed to generate more than $1 billion in revenues in the roughly contiguous EMEA (Europe, the Middle East, and Africa) region.
Although the non-Intel server segment includes mainframes, Unix deserves a big share of the blame for this drop-off. After all, we’ve already noted that Big Iron sales recovered in Q3 thanks to the availability of Big Blue’s new zEnterprise 196 mainframe systems.
It’s part of a depressing trend. Not counting the first quarter of 2009 -- during which Unix servers turned in an anomalously strong showing -- Unix has ceded share in every new quarterly tally. In Gartner’s Q3 tally last year, Unix generated close to $2.5 billion in revenues; this year, it accounted for $2.35 billion.
By contrast, Intel-based servers recorded more than $8 billion in Q3 sales.
Slowly, incrementally -- perhaps inevitably -- Unix is bleeding share. Will it fail to generate $2 billion in sales by this time next year?
Other Observations
If -- as we reported in last year’s recap -- the rubber met the road on the virtualization front, 2010 was the year in which virtualization consolidated its gains.
It might likewise have been the year in which cloud computing -- long hyped as the bee’s knees in transformative computing paradigms -- began to fulfill, in fits and starts, its promise. Microsoft, for example, shipped its Windows Azure cloud computing platform in February 2010. As history demonstrates, whenever Microsoft decides to throw its hat into the ring, a technology space has usually arrived. Nor does Windows Azure seem like a milquetoast offering, either: in a March address to students and faculty at the University of Washington, CEO Steve Ballmer pledged that “for the cloud, we’re all in.”
In 2010 we saw big hardware OEMs -- EMC Corp., IBM Corp., and Oracle Corp. chief among them -- get data analytics religion. EMC bought analytic database specialist Greenplum Software Inc.; IBM acquired analytic database pioneer Netezza Inc.; Oracle continued to enhance its Exadata analytic database platform. Meanwhile, Microsoft delivered its long-awaited analytic database offering -- SQL Server 2008 R2 Parallel Data Warehouse -- and promoted it in tandem with hardware partner HP.
The common strategy, according to industry watchers? A belief that “Big Data” analytics could drive beaucoup server, storage, and infrastructure sales.
Check back with us next December to find out if these and other bets panned out.