THE BUSINESS VALUE OF TECHNOLOGY

IBM and antitrust nonsense 12 | Teradata’s up to something good 16 | Your SaaS rights 20 Get firewalls under control 41 | Tech patents: alternative approaches 52

OCT. 26, 2009

[Plus]SHAREPOINT’S

SECOND ACTNow it’s taking on the Web p.14

SERVER BRAWL?Cisco looks ready to fight p.50

FCC Chairman Genachowski issues a dire warning

about wireless networks. Prepare now in case he’s right.

By Peter Rysavy p.23

‘Spectrum criSiS’

A United Business Media Publication® CAN $5.95, US $4.95

informationweek.com

THE BUSINESS VALUE OF TECHNOLOGY

Copyright 2009 United Business Media LLC. Important Note: This PDF is provided solely as a reader service. It is not intended for reproduction or public distribution. For article reprints, e-prints and permissions please contact: Wright’s Reprints, 1-877-652-5295 / ubmreprints@wrightsreprints.com

Oct. 26, 2009 23informationweek.com

[COVER STORY]

ulius Genachowski, chairman of theFederal Communications Commission,is worried. Not today, or even this year,but soon, he thinks the United Stateswon’t be able to meet the demand for

broadband wireless data. “The biggest threatto the future of mobile in America is thelooming spectrum crisis,” he told a wirelessconference earlier this month.

You should be worried, too—or, at theleast, be paying attention. Mobile datawill be an increasingly important piece

of the enterprise IT architecture, and alsopart of more products that companies pro-vide. CIOs need to identify network-centrictrouble spots today, as new wireless oppor-tunities arise and well before anything re-sembling a “crisis” hits.

Blame the iPhone. The first widespreaduse of wireless data began with the RIMBlackBerry in the mid-1990s, but it’s theiPhone that’s made wireless data a mass-market phenomenon. It’s changing whatpeople—employees and customers—ex-pect to be able to do with their smart-phones and how they interact with com-panies. This one blockbuster device hasgenerated enough of a network loadthat iPhone owners complain about re-sulting poor service, includingdropped calls, delayed messages, andslow download speeds.

Yet it’s more than smartphone apps

By Peter Rysavy

As wireless becomes a biggerpart of enterprise architectures,CIOs must understand the limits of today’s networks

Spectrum Crisis?

JJ

Spectrum Crisis?

24 Oct. 26, 2009 informationweek.com

[COVER STORY] WIRELESS DATA

that should have CIOs thinking aboutwireless spectrum. Sprint this monthcreated a business unit to sell use of itsspectrum for devices other thanphones to share data. Those devices in-clude anything from Amazon’s Kindlebook reader, to a water meter thatsends usage data, to a cash register (seestory, p. 26).

But there are limitations to today’swireless networks, and they’re going toget bigger. Don’t count on so-called 4Gtechnologies to save the day, at leastnot for many years. Until then, CIOsmust understand the effect on applica-tions when the network gets over-loaded, and plan accordingly.

Demands On The NetworkA good place to start is the actual de-

mand compared with network capac-ity. Cisco, in a January report, predictedthat mobile data traffic will more thandouble every year through 2013, atwhich time it will reach 2 exabytes—2 billion GB—per month. Think Ciscobelieves its own numbers? This month,it struck a $2.9 billion deal to buy Star-ent Networks, whose equipment helpswireless carriers deliver multimedia andother data-intensive features.

The wireless industry is petitioningthe FCC to allocate additional spec-trum over the next six years to morethan double the industry’s licensedspectrum. FCC Chairman Gena-chowski sounds sympathetic. “Evenwith innovative spectrum policies andinnovative new technologies, expertsbelieve we are way too likely to becaught short,” he said at the CTIA con-ference this month.

IT managers should care becausecongested networks are slow and,worse, unpredictable. As throughputrates decrease and packet delays in-crease, some applications start behav-ing erratically, as we’ll discuss.

It’s not panic time. We haven’t yetreached widespread saturation on to-day’s 3G networks, but given currentmarket growth, it will likely happen

without the timely addition of new spec-trum. And given the multiyear processfor identifying, auctioning, and deploy-ing new wireless spectrum, spectrummight not become available fast enoughto accommodate the rising tide of smart-phones and data-hogging applications.The Cisco report states that smartphonesgenerate more than 30 times the trafficof basic cell phones; laptops with wire-less modems 450 times the traffic.

Operator StrategiesData traffic now represents 25% of

the revenues of wireless network oper-ators globally, and it’s a more profitableand faster-growing business than voice,so operators such as AT&T and Veri-zon are highly motivated to find tech-nological ways to squeeze data capacityout of their existing networks.

One is to deploy new networks inspectrum that they already own. BothAT&T and Verizon have spectrum inthe 700-MHz band, previously usedfor UHF TV channels, where they willdeploy a technology called Long TermEvolution. Verizon, for instance, plansto start deploying LTE next year andAT&T in 2011.

However, LTE isn’t enough. While itcould more than double operators’current mobile broadband capacity, itwon’t be fully deployed until about2014. The operator in the strongestspectrum position right now to pro-vide mobile broadband, and the onewith the least amount of fine printabout what customers can do on its

network, is Clearwire and its WiMaxnetwork. That network, however, alsohas the smallest footprint since deploy-ment began only recently.

Operators also are upgrading 2G sys-tems to 3G LTE to wring more capacityout of the same spectrum. Over time,probably sometime in the middle of thenext decade, these same 3G networkswill be upgraded to some more highlyevolved versions of LTE or WiMax,what’s envisioned as 4G. But near term,unfortunately, wireless technologies areapproaching their theoretical limits, lim-ited by what’s called the ShannonBound, which dictates the maximumspectral efficiency relative to noise.

Beyond spectrum and efficiency, op-erators also can increase the number ofbase stations. Smaller coverage areasequate to more available spectrum perperson, the reason Wi-Fi can offer suchhigh throughputs. Operators not onlyare deploying more large-area cells,called macrocells, but they’re also de-ploying more microcells and picocellsthat cover a highly localized area—forexample, a shopping mall.

A promising option in its earlieststages is femtocells, which cover verysmall areas and use ultra-low power viaaccess points in homes or businessesthat plug into subscribers’ broadbandInternet connections. Operators seefemtocells as a great way to off-loaddata onto wired networks when peopleare in their homes or businesses.

Likewise, you may have noticed op-erators encouraging Wi-Fi use. Let-

Enterprise Mobility How companies are pushing to mobilize businessapplications: informationweek.com/analytic/mobileapps

LTE Vs. WiMax This won’t be a typical winner-take-all duel:informationweek.com/analytic/lte

802.11v Ahead Upcoming standard could helpcompanies get a grip on wireless network usage:informationweek.com/analytic/80211v

DIG DEEPER

See all our reports at analytics.informationweek.com

26 Oct. 26, 2009 informationweek.com

[COVER STORY] WIRELESS DATA

ting 3G subscribers use hotspots freeisn’t benevolence; operators are tryingto off-load as much data as they canonto Wi-Fi.

Finally, one of the most importantdials operators can turn to control vol-ume is pricing. Wireless plans for lap-top connectivity remain high, at about$60 per month. Smartphone plans arelower, at $30, since phone usage usu-ally consumes less data, but the iPhoneis challenging that assumption. Use aphone as a modem and the operatorwill promptly require the higher-pricedlaptop plan. There is also the somewhatcontroversial 5-GB monthly cap onmost U.S. laptop plans. Business tech-nology managers should be encouragedthat operators are so active in manag-ing demand and in continuing to invest

in their networks. Wireless networksoften work reasonably well. But thetoughest issue is lack of predictability.It’s problematic as companies pushmore applications over smartphones tocritical mobile workers such as servicepersonnel and salespeople.

Enterprise ImplicationsAn employee working with a wire-

less application once a day, perhapsuploading or downloading a batchfile, may not care too much about de-lays or retries. But a worker with aproductivity application accessing thenetwork continually throughout theday needs a dependable system. Thiscan be achieved, but not without tar-geted effort, and generally only withapplications that were specifically de-

signed for mobile connectivity. Though defined in wireless specifi-

cations, no wireless operators todayoffer quality-of-service guarantees,such as guaranteed bit rates, maxi-mum latency, or traffic priority. Soperformance is purely best efforts.U.S. operators quote typical through-put rates, which means that some rea-sonable percentage of users—perhaps80% or even 90%—will experiencethat range. Business technology man-agers must realize that leaves as muchas one in five users out of the typicalrange—higher or lower—with somegetting relatively low throughput.

Most TCP/IP-based networking ap-plications weren’t made to operate overwireless connections. While today’s 3Gand tomorrow’s 4G networks can de-

Imagine running an entire retail store location over a 3Gwireless network—including transferring inventory up-dates, point-of-sale data, and payroll information. FollettCorp., which operates 900 college campus bookstores

across the United States,has proved it’s not only possible,but ithas been a reliable stopgap for the few months it takes Follettto hardwire a new bookstore.

Follett’s store locations can vary widely: Some are store-fronts on a city street; others take up a floor inside a studentunion building. After signing a lease for a new store, it cantake two to three months for Follett to configure a conven-tional frame relay network between the store and the com-pany’s home office. 3G, in some locations, is the remedy.“It’sallowed us to have any one of 33 applications functional onday one,” says Roger Laframboise, Follett’s director of net-work services.

Yet the approach works only with some new stores, giventhe inconsistency of cellular service providers’ coverage.Stores in metropolitan areas are the best candidates. 3G testsin stores away from big cities haven’t been as successful—re-sulting in lost connections in the middle of cash-register ring-ups, for example—for the same reason people have spottycell service in certain regions: fewer cell towers.

“In a test store right outside of Chicago, the performancewas fantastic,”says Bob Richard, manager of planning and ad-

ministration in Follett’s IT group.“When it’s working right, it’s asfast as a traditional relay.” In fact, in an ongoing experiment,two stores have been running entirely on a 3G network formore than a year, supporting such tasks as staff scheduling, in-ventory management, and cash and credit card transactions.Follett also prioritizes traffic across a 3G network based on itsimportance—revenue-generating transactions, for example,take precedence over inventory management.

Laframboise advises companies to have contracts with sev-eral 3G carriers before trying this kind of effort,since carrier cov-erage varies at different locations.Any number of factors—loca-tion of wireless towers, the technology carriers use, buildingconstruction materials,and even typical weather patterns—canaffect coverage and signal strength. IT teams should plan to doan intensive site survey and assessment to pick the right car-rier, including assessing whether 3G is even an option.

While wireless has proved itself for quick retail launches,Laframboise says it has a long way to go before he’d use it asthe permanent network for critical enterprise apps. There’s ajumble of standards in the United States alone—includingGSM and CDMA, with EV-DO, HSPA, WiMax, and, soon LTE—and neither CDMA nor EV-DO mesh with European and Asianstandards.That suggests caution before building a long-termenterprise apps platform on wireless broadband.

—Mary Hayes Weier (mhayes@techweb.com)

Follett Runs Some Stores On 3G NetworksSPEED TO MARKET

28 Oct. 26, 2009 informationweek.com

[COVER STORY] WIRELESS DATA

liver IP packets reliably and efficientlyin a congested situation, or even withjust a very weak radio signal, through-put rates can decrease significantly, de-lays can increase, packets can drop, andconnections can be lost entirely. Gettingreconnected might require a differentIP address, which can confuse an ap-plication in midtransaction. Movingrapidly in a train or car also can stressthe connection since the quality of theradio signal fluctuates widely.

There are ways to minimize the painof such difficulties. Those include back-ground synchronization so users don’thave to wait as the application commu-nicates with the server; longer time-outs

to tolerate wireless network variations;restarting previous data exchanges fromthe point of failure; and local cachingand data compression to minimize theamount of data communicated.

To get those kind of robust commu-nication methods is to build an appli-cation from the start to run over awireless network, the way wireless e-mail on RIM’s BlackBerry or the thou-sands of iPhone apps were. But forlegacy apps, mobile middleware is usu-ally the answer.

Some mobile middleware can be in-stalled to work with existing applica-tions, such as mobile VPNs from com-panies such as NetMotion Wireless.Others, from companies such as An-tenna or Sybase, provide a develop-ment environment with which to buildwireless-optimized applications. Bene-fits of middleware include not only themore robust communications trans-ports mentioned above, but with somepackages, the ability to target multiplesmartphone platforms with the sameapplication code base.

Middleware, however, comes withtrade-offs. It adds to the cost and sys-tem complexity, and it involves alearning curve for IT departments andintegrators. And any applications de-veloped using the middleware tend tobe tightly bound with it, making it dif-ficult to change middleware vendorsdown the road.

IT managers must understand whatmakes some applications inherently bet-ter for wireless than others. For example,SQL-based database applications can in-volve a large number of SQL statementstraversing the air, and with wireless la-tency much higher than wireline, thiscan translate to very sluggish perform-

ance. On the other hand, a database mo-bility extension via middleware, or fromthe database companies themselves, gen-erally involves an architecture where themobile system uses efficient and resilientprotocols to communicate with a mobileserver, which then acts as a proxy forcommunication with the end server.

With the move to cloud computing,IT managers may wonder if Web appli-cations will take care of many of theseproblems. Unlike native applications,Web apps are highly dependent onsolid network performance for a gooduser experience. On the other hand,Web interactions are relatively state-less, meaning TCP connections don’tremain open after the browser accessesa page. So network interruptions be-tween page loads shouldn’t affect usersessions. The result is that pages mayload more slowly in the event of net-work congestion, but overall, sessionsshould be fairly resilient. Of course, inreally congested situations, page loadsthemselves will time out.

Applications that require constant

high bandwidth are particularly vulner-able to disruption and thus a poor fitfor wireless delivery today. A file down-load can tolerate variation in through-put fairly well, but streaming video(e.g., training, a repair procedure) ismore susceptible, especially if the videoviewer doesn’t buffer data. Audiostreaming, though, is practical—radiostation streaming is very popular on theiPhone. Bottom line, the more real-timeand high-bandwidth the app is, themore vulnerable it is.

IT managers really have two funda-mental approaches available to them.For lightly used or noncritical applica-tions, standard apps that aren’t opti-mized for wireless probably will worksufficiently well to be viable. For criti-cal applications, however, IT managersshould consider optimizing their appsfor wireless.

Field testing in representative cover-age areas also is a must. Given variationsin network performance, tests shouldbe done across multiple iterations, inmultiple locations, and during differenttimes of the day. While there’s no ruleof thumb, testing projects I have con-ducted used six to 10 iterations of eachapplication test case in at least six to 10locations in a single metropolitan areato model expected performance. Usu-ally, it’s a good idea to test in more thanone city if the application will be usedin multiple cities. Testing should bewith both 2G and 3G connections.

Where Things Are HeadingNetworks will get faster and capacity

will increase, but neither will necessar-ily keep up with escalating demand. Sothis state of affairs will remain with usfor a long time to come. There areplenty of great reasons to use mobilebroadband to improve productivity. Butradio will never be the same as wire.

In the short term, IT managers shouldmonitor the net neutrality debate, whichhas recently shot back into the politicalagenda. The strictest interpretation ofnet neutrality, for example, would effec-

IT managers must understand whatmakes some applications inherentlybetter for wireless than others.

tively let any user do anything on thenetwork, no matter how bandwidth-in-tensive. That would let businesses exper-iment with high-bandwidth uses butcould also prevent operators from im-plementing quality-of-service guaran-tees, since QoS can be used to give somepackets higher priority than others.

These issues are being hotly debated.Some net neutrality backers say quality-of-service guarantees can be accommo-dated within the reasonable limits of netneutrality regulation. The wireless in-dustry is lobbying hard for measuressuch as network management thatwould let them throttle back users whoare making excessive demands on thenetwork. The resolution of these issuescould affect how predictable networkperformance ends up.

Another problem to anticipate isthat, as networks get faster, the varia-tion in throughput will increase. This

is because technologies such as LTEand WiMax can exploit good radioconditions for extremely high through-puts, using high-order modulation andadvanced antenna techniques. Butwith a weak signal, they have to fallback significantly in throughput to getdata across reliably.

A report by this author for 3GAmericas, a 3G advocacy group forHSPA/LTE technologies that includesvendors and operators, projected thatLTE speeds will range from 2 Mbpsto 12 Mbps in typical 10-MHz de-ployments. Should a user be out of ametro area on a 2G network likeEDGE, throughputs could be as lowas 100 Kbps. Not always having ac-cess to the fastest technology in alldesired coverage areas will remain acommon condition indefinitely. Eventoday, 3G isn’t ubiquitous despite ini-tial deployments that started at the

beginning of the decade.One area that may bring improve-

ment is quality-of-service agreements,assuming operators decide to and canmake them available. Technologiessuch as LTE can readily provide differ-ent classes of service, including trafficprioritization and even guaranteed bitrates. But that will all take time andregulation, and even then you couldroam onto a partner network, or 3G,or even 2G networks, where the sameoptions won’t exist.

So IT managers should take mattersinto their own hands now and deploysystems and applications with realisticexpectations of performance, and ap-propriate protective measures.

Peter Rysavy is the president of RysavyResearch (www.rysavy.com), a consult-ing firm specializing in wireless. Writeto us at iweekletters@techweb.com.

[COVER STORY] WIRELESS DATA