`
`1998 NovemBrr/DECEMBER ONLINE 89
`
`>
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page 1 of 8
`LGEv. Uniloc
`
`
`
`There’s a new I
`Unlike green-
`machine packsa |
`big, boxy “tower”
`a slim, futuristic |
`performs muchlikea |
`tions drastically unlike
`
`by Mark Sheeh
`
`
`
`because they do no longer:
`cal
`storage.‘Some of
`them havelittle more memory than ittakes to paint
`a video screen because that’stheironly job, A few of
`them contain processors—286sfor example—more
`familiar in computer museums thaniinthecorporate
`cubicle maze.
`
`
`
`SOME BACKGROUND
`The thin client story has its roots
`in client/server technology. In the
`beginning was the mainframe, a
`powerful computer for its day with
`access limited to a few elite pro-
`grammers. As mainframesgot easier
`to use and more of us began using
`them, the waste and hassle of the
`mainframe’s paper tape and punch
`card interface became vexing. In
`response, the terminal was invented.
`The mainframestill did all the
`processing, but the terminal took
`care of the user’s input and output.
`(Of course, a lot of output was
`diverted to the printer, too.) The
`expectation of access to the main-
`frame via terminal from remote
`offices or from home brought us the
`beginnings of computer networks.
`Terminals started out “dumb.”
`They put letters and numbers on the
`screen in fixed rows and columns.
`
`|ICA—Independent Computer
`Architecture
`
`NC—Network Computer
`
`RDP—Remote Desktop Protocol
`
` Guide to Acronyms
`
`TCO—Total Cost of Ownership
`
`WT—Windows Terminal
`
`buildings, distant towns, and even-
`tually distant countries.
`To unify network users around a
`fountain of functionality, the server
`computer was born. Early servers
`gave scattered PCs access to central
`printers and files. They intermedi-
`ated, giving and denying access,
`providing security and backup...and,
`
`
`
`90 http://www.onlineinc.com/onlinemag
`
`...we began to resent the need to tie our
`terminals to mainframes through copper
`umbilical cords and the industry gave us the PC.
`
`Writing in Spanish? Sorry, no accent
`marks or upside down question
`marks available. Writing equations?
`Sorry, no math symbols available, let
`alone super or subscripting. Drawing
`pictures? Unthinkable!
`Later, “smart” terminals overcame
`these obstacles. Soon thereafter,
`with our expectations raised and our
`individuality beginning to assert
`itself, we began to resent the need to
`tie our terminals to mainframes
`through copper umbilical cords and
`the industry gave us the PC. Suddenly
`the mainframe was obsolete as a
`general purpose computer. Salesfell
`off as the information producing and
`consuming markets tooled up to create
`islands of desktop automation—
`islands of PCs.
`Immediately, of course—because we
`love to socialize almost as much as we
`love our independence, and because
`our managers love to economize—
`there sprang up a hundred-headed
`Hydra of network technologies
`aiming to link these isolated PCs
`back together. Local area networks
`sprang up withinoffices and build-
`ings. Wide area networks sprang
`up too, linking PCs in distant
`
`incidentally, creating one more point
`of failure. See “A Pig Under the
`Hearthrug: The So-Called Hidden
`Costs of PCs” sidebar.
`All this time, the mainframe
`continued to be a force in data
`processing. As mainframes scaled
`down into economical mini versions,
`and as network servers scaled up to
`do more mainframe kinds of things,
`the distinction blurred. It was fuzzed
`further by the fact that many kinds
`of special-purpose personal comput-
`ers had become as powerful as most
`servers and some mainframes!
`Mainframesof whatever size contin-
`ued to do what mainframes had
`always done. They served as
`database repositories. They hosted
`data too sensitive or software too
`complex or too expensive to run on a
`PC. Throughout the history of the
`PC, terminal emulation software was
`a standard component; it was hard to
`buy a $2,000 PC without also receiv-
`ing software to turn it into a $200
`terminal.
`
`CLIENT/SERVER COMPUTING
`Well this relationship between
`PCs and mainframes was shameful!
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page2 of 8
`
`
`
`While networks had been importantsince the
`invention of terminals, they became absolutely
`pivotal in the client/server model of computing.
`
`My manager in the mid-80s used to
`wail, “Gigaflops of processing power
`on thousands of enterprise desktops
`being used for what? To emulate
`dumb terminals!” To improve perfor-
`mance and to make better use of the
`enterprise’s investment in PCs,
`client/server computing came into
`being. The PC wastheclient and the
`mainframe—or something like
`it—wasthe server. Theclient had its
`own special software: software that
`crunched numbers, for example, or
`stored and managed a personal elec-
`tronic mailbox. And the server had
`its special software too: software
`that managed multiple clients’
`access to a database of numbers, or
`that received incoming mail messages
`for a hundred users and held them
`until the client software cameto fetch
`them. With client/server computing,
`the PC and the mainframe entered
`into a partnership, each doing what
`it did best: the PC interacting with
`the user, storing personal files,
`controlling local printers; and the
`mainframe managing data, storing
`sharedfiles, and controlling shared
`peripherals. Linking the two part-
`ners was a data network. The more
`active the partnership, the bigger the
`data pipe needed to link the client
`and server, While networks had been
`important since the invention of
`terminals, they became absolutely
`pivotal in the client/server model of
`computing.
`Mostclient software nowadaysis
`written for “fat” or (my preference)
`“thick” client computers. In many
`cases, the thicker the client machine
`is, the better. At Montana State
`University, for example, we are
`implementing SCT’s Banner2000
`suite of administrative software
`packages. The desktop client computer
`SCT recommends for the average
`user is a 233MHz Pentium II with
`32MBof memory, a 3.2GBhard disk,
`and a 100Mb Ethernet adapter.
`(This is a very thick client drinking
`data througha fire hose!) In an envi-
`ronment that has been based on
`cheap, dumb terminals connected to
`
`switched on, it sends a boot request
`to the server. The server responds by
`downloading the operating system
`into the NC over the network. (This
`can be a long process, by the way,
`even over a fast network.) A graphics-
`based desktop appears, and offers
`the user a choice of applications. The
`server then downloads the applica-
`tion the user selected to the NC’s
`memory, whereit executes. Data files
`or portions of data files—text docu-
`ments, spreadsheets, databases—
`may also be downloaded as the
`application does its stuff. When the
`user’s task is completed, the data
`files are saved on the server and the
`NC’s memory becomesavailable for
`another application.
`The applications that NCs run are
`of two basic kinds: native applica-
`tions written for the NC’s processor,
`or Java applications written to run
`under Java Virtual Machinesoft-
`ware that is downloaded to the NC
`over the network. Java applications
`are by far the more common of the
`two types. Once written they can
`run on a variety of NCs. Native
`applications, by contrast, must be
`created in a different version for
`
`mainframes, the move to replace
`every desktop terminal with a thick
`client PC is a multimillion-dollar
`proposition,
`MSU is just a handy local
`example. The same problem occurs
`everywhere. Terminals have to be
`replaced with $1,500 PCs. Existing
`PCs have to be replaced with bigger,
`faster, “thicker” ones. Expensive
`network connections need to be
`upgraded to meet the application's
`need for speed. Total cost of owner-
`ship (TCO) figures are genuinely
`frightening.
`
`BUT, SOFT! WHAT LIGHT
`THROUGH YONDER WINDOW
`BREAKS?
`Somesee a glimmerof hope in the
`thin client. Stephen Bell introduced
`ONLINEreaders to thin client tech-
`nology last year [2]. Just to recap, a
`thin client is a kind of minimalist
`computer. There are two basic types:
`
`Whenthe network computeris switched on,it
`sends a boot request to the server. The server
`responds by downloading the operating system
`into the NC over the network.
`
`the network computer (NC) and the
`Windows terminal (WT). They’re so
`different that each deserves a
`description of its own. A third type
`of thin client, the NetPC, is no longer
`a player in the marketplace [3].
`Because NCs and WTs require
`networks to function, their kind of
`computing is referred to as “network-
`centric” computing.
`
`Network Computers
`Network computers (NCs) are
`little more than a processor chip,
`some memory, a screen, a mouse,
`and a keyboard. They connect
`through a network cable to a server
`computer, The server houses the
`thin client’s operating system and
`application software, as well as the
`files belonging to the user and the
`user’s workgroup. When the NCis
`
`each type of NC processor. Hence,
`they are more expensive for manu-
`facturers to create and maintain and
`are less flexible for server managers
`to deploy.
`Most NCsare very thin indeed
`with regard to disk resources. But as
`the NC concept has matured, the
`network, memory, and processor
`requirements to run fully featured
`Java applications have grown out of
`hand. At least one manufacturer’s
`NC, configured with its optional
`hard disk drive, is physically distin-
`guishable from a classical thick
`client only by its lack of internal
`expansion ports (and definitely not
`by its price!),
`Manufacturers of NCs include Sun
`Microsystems (JavaStations), IBM
`and its partner NCD (IBM Network
`Station, NCD Thinstar), Neoware
`
`>
`
`1998 NOVEMBER/DECEMBER ONLINE91
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page 3 of 8
`
`
`
`
`
`en
`
`
`
`...the WT client simply sends keystrokes and
`mouse movements to theserver...
`The programs...execute entirely on the server,
`using the server’s processor, memory, anddisk.
`
`(NeoStations), and Acorn Computers
`Ltd. (Acorn Corporate NC)[4].
`
`Windows Terminals
`A completely different model of thin
`client computing is embodied in the
`Windowsterminal (WT). Rather than
`downloading everything from a server
`and running it locally, the WT client
`simply sends keystrokes and mouse
`
`movements to the server. The client
`then displays what the server does
`with those inputs. It does no processing
`other than that needed to paint the
`screen. The programs a WT launches
`on the Windowsserver are ordinary
`Windows programs (newer 32-bit
`programs work best). They execute
`entirely on the server, using the
`server's processor, memory, and disk.
`
`RDP vs. ICA
`
`Server
`
`While WTs are certainly thin, in
`the sense that they take almost no
`active role in running Windows
`applications, there is no standard
`hardware configuration for WTs, as
`there is for NCs. WT hardware can
`range from something very much
`like a traditional terminal to a
`powerful UNIX or NT workstation.
`NCs can be set up to behavelike
`WTs, too. See the sidebar “Wait!
`You’re Both right!”
`The onestrict requirement for WTs
`is that the terminal device support
`either Microsoft’s remote desktop
`protocol (RDP) or the independent
`computer architecture (ICA) protocol.
`RDP or ICA client software can be
`brought to bear in three ways. It can
`reside on a chip inside the terminal,
`as it does in the new, dedicated WT
`products and some NCs;it can be
`downloaded into the memory of an
`NCand executed there; or it can be
`loaded onto the disk of a PC,
`Macintosh, or UNIX workstation for
`execution on demand.
`RDP is used when the WT is
`running a recent version of a Micro-
`soft operating system (Windows 95
`and 98, Windows NT Workstation,
`Windows 3.11) or when the terminal
`has been designed specifically to
`support RDP (as some NCs anddedi-
`cated WTs do), ICA is used for client
`machines running DOS, UNIX,or the
`Macintosh operating system.
`RDPand ICAserver software runs
`on a dedicated Windows server
`
`RDPClient
`
`
`
`
`
` edt
`teMi
`
`=
`
`ICA Client
`
`
`
`PC running:
`* Windows 3.x, Windows 95/98, or
`Windows NT client
`
`* Some network computers
`* Some Windows terminals
`
`Microsoft NT server running:
`* NT vA Terminal Server Edition, or
`* NT v3 with Citrix WinFrame for RDP
`clients and Citrix MetaFrame for ICA
`clients
`
`92 http://www.onlineinc.com/onlinemag
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page4 of 8
`
`
`
`
`
`Windows95 taskbarto clients not
`running Windows95.
`Microsoft considers the top five
`vendors of WTs to be Wyse (Winterm
`series), Tektronix (ThinStream),
`Boundless (Viewpoint series), NCD
`(Thinstar), and Neoware (@worksta-
`tion series) [6].
`
`WTs are in the sameprice range, but
`include monitors.
`More significantly, NCs and dedi-
`cated WTs are said to be muchless
`expensive to manage and support
`than PCs. In spring 1996, Zona
`Research predicted a five-year, total
`cost of ownership savings of 57% for
`Wyse Windows terminals versus PCs
`(7]. In summer 1997, Microsoft
`suggested that savings of 46% in TCO
`were possible in the same scenario[6].
`In late 1997 Gartner Group predicted
`a 22% savings for NCs versus PCs[1].
`While there’s wide variation in these
`numbers, the trendis clearly positive.
`Whyis this so?
`The greatest virtue NCs and WTs
`have in the management and
`support arena is that their software
`environments are stored entirely on
`centrally controlled servers. Because
`most NCs have no disk drives, the
`average NC user can’t bring soft-
`ware or data from homeontothe job
`and thus can’t introduce viruses or
`other system complications that are
`machine. This server’s only job is to
`expensive to combat. Nor can they
`serve WTs, offering a selection of
`download software (other than Java
`Windowsapplications to all WTs
`Applets) from the Internet and run
`authorized to use them. Because
`it on their desktop computers with
`these servers run Windowsapplica-
`the attendant risks of viruses and
`tions for their client machines, they
`system software conflicts. Dedicated
`must be Intel or Intel-like computers.
`WTs are similarly constrained.
`The server software that supports
`With NCs and WTs, when the
`RDPaccess under Microsoft Windows
`enterprise decides to upgrade to the
`COSTS AND BENEFITS
`NT is called Microsoft Windows NT
`next version of a software package,
`NCs and WTs can both address
`4.0, Terminal Server Edition. The
`the change occurs in one place: on
`the high total cost of PC ownership.
`server software that supports RDP
`the server. There’s no need to visit
`This is less true if we use Citrix
`under earlier versions of NT is called
`every desktop to install the software
`MetaFrame to make WTs out of
`Citrix WinFrame.
`(and then visit it again and again to
`Macintoshes, fully configured new
`Citrix also supplies add-on NT
`figure out why the new version
`or old PCs, or UNIX workstations,
`server software, called MetaFrame,
`doesn’t work with that particular
`because those machines can be
`which supports ICA access. (For a
`set of hardware and software).
`tinkered with more by their owners—
`while, Insignia Solutions, Inc. sold an
`Because NCs and WTsare used
`thought to be a major TCO factor in
`ICA-based product called NTRIGUE!
`with a controlled, limited set of soft-
`desktop computing. But if we stick to
`that delivered Windows applications
`ware, training and end-user support
`to NCs and other terminal devices, In
`NCs as NCs or if our WTs are simple,
`requirements are reduced. And
`limited devices such as NCsor dedi-
`the first quarter of 1998, Insignia
`because they are fairly idiot proof,
`cated WTs, they can definitely help us
`sold its NTRIGUE! Technologies to
`these devices are hard for end users
`minimize ownershipcosts.
`Citrix Systems,Inc.)
`to mess up [8]. It’s difficult to over-
`First, NCs and dedicated WTs are
`MetaFramealso provides such
`emphasize this benefit. In higher
`less expensive to purchase than most
`functions as load-balancing when a
`education [9], and almost every-
`PCs. NCs cost from $500 toalittle
`large farm of NTs is set up to share
`where else computers are used, ways
`over $1,000 plus monitor. Dedicated
`several NT servers, and provides a
`to reduce the cost of support for indi-
`vidually managed PC environments
`have been the subject of endless
`debate.
`Another common benefit of NCs
`and WTsis that they help us get
`aroundthe obsolescence issue. Many
`
`The greatest virtue NCs and WTs havein the
`managementand support arenais that their
`software environments are stored entirely on
`centrally controlled servers.
`
`94 http://www.onlineinc.com/onlinemag
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page 5 of 8
`
`
`
`Theindustry just seems ur
`
`of us intuitively expect an invest-
`mentin technology—say $2,000 for a
`desktop computer—to havea life-
`time of many years. Most of us have
`come to understand, however(if not
`to accept), that the computer we buy
`this year is technologically obsolete
`in a few monthsandis functionally
`obsolete within two years. Citrix
`claims that in theory its MetaFrame
`product can make a fully functional
`WT out of a 14-year-old 80286
`computer(like the 1984 IBM PC-AT).
`
`At the practical level, according to
`Doug Palmer of DataSource Hagen,
`a Minneapolis area NCreseller,
`running even the basic Windows
`terminal software on anythingless
`than an 80386 SX is painful. Either
`way, it appears that machines at
`least ten years old can function
`usefully delivering 1998 Windows
`applications in Windows terminal
`mode. For those of us who have
`upgraded our computers several
`times since 1988, this is good news
`indeed.
`Most IT managers havefound it
`difficult to build desktop computer
`depreciation into institutional
`budgets in a way that provides for
`the two-to-four year replacement
`cycle we're locked into with desktop
`PCs and their software. The WT
`concept looks like a way out of that
`cycle. NCs used as WTs should carry
`the same immunity to obsolescence.
`
`COMMON DISADVANTAGES
`I think the TCO benefits of central-
`ly served applications are overstated,
`whether we're serving Java applica-
`tions to an NC or Windowsapplica-
`tions to a WT. This is largely just an
`opinion, but I can present twolines of
`evidence to support it. First, I’ve tried
`to hire NT and UNIX system admin-
`istrators for the past several years,
`and I can tell you they’re very hard to
`find and very expensive to retain. You
`need a good system administrator or
`two if you’re going to deploy applica-
`tions in a network-centric model.
`Networks of modern terminals and
`NGs are complex environments. True,
`each individual computer owner can
`forget about managing his or her own
`
`
`
`
`computer. But to use NCs or WTs
`you'll need to hire server support
`staff who really know theirstuff.
`Second, NC servers have a “dirty
`little secret” [10] that I suspect may
`be shared by WT servers: they are
`bears to set up! As we evaluated NCs
`at Montana State, we set up server
`environments for Sun, IBM, and
`Neoware NCs.All were difficult, each
`in its own way. Part of the problem is
`that the technology is young. We were
`constantly applying vendor-supplied
`patches to the servers to add crucial
`features and fix minor bugs. The
`vendors’ paper documentation was
`generally outdated and was nonexis-
`tent in some cases. The online docu-
`mentation was often very much in
`flux, sometimes changing daily. In
`mid-1998, it is difficult to obtain eval-
`uation units of NCs for these
`reasons—the vendors know that the
`machines will make a bad impression
`on the technicians installing them.
`
`The industry just seems unprepared
`to market NCs. Perhaps because
`they have been rushed into produc-
`tion in a frenzy of competition, most
`of these devices seem morelike
`“projects” than “products.” Know-
`ledgeable technical help is very hard
`to find. People whoare installing the
`systems in the field must do a lot of
`research, they must learn new tech-
`nologies unnecessarily to supple-
`ment the gaps in the vendors’
`technical support organizations, and
`in general they finish their evalua-
`tions with a bad taste in their
`mouths. By the time you read this
`article the situation may have
`improved immensely. Other vendors
`may be better or worse than those
`we worked with. But if one factor is
`impeding the widespread adoption of
`NCs, I would sayit’s this one.
`Windowsterminal server software
`is being marketed more effectively. If
`you point your browser to Citrix’s
`
`>
`
`1998 NOVEMBER/DECEMBER ONLINE 95
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page 6 of 8
`
`
`
`Web site, http://www.citrix.com/
`free.asp in particular, you can down-
`load a plug-in for your Web browser
`that will allow you to test-drive
`Windowsapplications served by
`WinFrame. (You'll find these applica-
`tions in the “Demo Room” at http://
`www.citrix.com/demoroom/default.
`asp.) The fact that the NC vendors
`I’ve worked with can’t or don’t
`provide a similar Internet-based
`server experience with their evalua-
`tion units is one of the reasons I
`think of NCsas projects rather than
`products. If I wanted to sell a lot of
`something I’d sure want to makeit
`easier to try out than today’s NCis.
`The organizations that predict low
`TCO for NCs and WTs undoubtedly
`figured in somecosts for the hassles
`of server upkeep. My experience
`with NCs suggests they may have
`substantially underestimated those
`costs.
`Another common drawback to NCs
`and WTs is that their server comput-
`ers represent single points of failure
`that may affect many users. With a
`local area network of PCs, if the
`network or the server goes down, at
`least some useful work can be
`accomplished by the computers
`running separately. With either NCs
`or WTs, if the network or the server
`goes down, we might as well take
`the rest of the day off. NCs can
`continue to limp along with the
`applications already downloaded,
`though no stored files can be opened,
`and no openfiles can be saved. Fully
`configured PCs acting as WTs can be
`used offline to the extent that they
`havelocal copies of useful software
`installed. Dedicated WTs simply stop
`working. Period.
`A last common drawback to both
`NCs and WTsis the cost of the
`servers and server software to
`support them. This is much less an
`expense for NC servers than for WT
`servers because an NC needs a
`server only at startup, when an
`application is being loaded, and
`whenfile or print operations need to
`occur. A WT uses terminal server
`resources every momentit is turned
`on. Fifty NCs would require a
`portion of a well-configured UNIX or
`NT server. The cost for such a server
`would be on the order of $5,000-
`$10,000. Oft-quoted standards for
`
`96 http://www.onlineinc.com/onlinemag
`
`Microsoft Windows NT 4.0, Terminal
`Server Edition include 32MB of
`memory for the system and 8MB of
`RAM for each user [11]. Doug Palmer
`of DataSource Hagen suggests 50
`users might be adequately served by
`a two-processor, 350MHzIntel server
`with 512MB of memory and 9-18GB
`of RAID disk. By my estimation,
`the cost of the hardware and NT
`operating system for such a server
`would be on the order of $25,000.
`The approximate cost of Citrix
`MetaFrame, to give Windows termi-
`nal access to 50 non-Windows PCs,
`would be an additional $13,000.
`Network capacity needs may also
`increase if you move from PCs on a
`LAN to network-intensive NCs or WTs.
`
`INDIVIDUAL COSTS AND
`BENEFITS
`NCs used as NCs (to run Java
`applications, for example) may be
`more subject to obsolescence than
`WTs, depending on the demands
`placed on them by the applications
`written for them. For example, during
`our evaluation of the IBM Network
`Station at Montana State University,
`we were forced to upgrade the
`
`A still clearer cost difference exists
`between server software products. NC
`server software is just an application
`that runs under ordinary operating
`system software. (IBM’s NC server
`software, however, currently requires
`a dedicated server if run under the
`NT operating system.)
`In the
`Windows terminal environment, on
`the other hand, Microsoft appears to
`be gravitating rapidly toward requir-
`ing the server to run the Terminal
`Server Edition of NT. Diminishing
`competition here means price breaks
`will be hard to comeby.
`
`WRAPPING UP
`Which approach to enterprise
`computing is right for your organiza-
`tion? The answeris complex. If your
`use of information technology tools
`is exploratory and innovative,
`neither NCs nor WTs makes much
`sense. Stick with your “thick” PCs.
`That’s the case for a lot of us in
`higher education andfor those in the
`research and creative activity organi-
`zations in government and industry.
`For those who do single-function jobs,
`however, whetherit’s word processing
`or database management or even
`
`For those whodosingle-function jobs...it might
`make sense to shift toward NCs or WTs.
`
`machine’s memory to 64MB to make
`room for the Lotus eSuite of general
`office applications. WTs appear likely
`to require muchless in the way of
`regular upgrades.
`NCsare likely to have access to a
`much narrowerselection of software.
`Many major vendors, including Lotus
`and Oracle, are developing Java
`versions of their applications for use
`on NCs and Java-enabled browsers.
`Limited software options should not.
`be a problem for WTs, as long as the
`Windows software market remains as
`competitive as it is now.
`Windowsterminals must have their
`applications seryed to them by an
`Intel-based NT server. While there
`are many vendors of such servers, a
`far wider variety of options is avail-
`able for NC servers. The greater
`competition here could make an NC
`server less expensive, user for user,
`than a WT server.
`
`Web surfing, it might make sense to
`shift toward NCs or WTs. The server
`support requirements of those envi-
`ronments make them ill-suited to
`very small businesses, at least until
`our local Internet service providers
`begin selling access to professionally
`managed, subscription-based NC and
`WTservers. Larger organizations,
`though, are finding that NCs and WT
`servers are saving them money in
`many ways. (Look for testimonial
`success stories in the promotional
`brochures of the suppliers listed in
`the “Internet Resources” sidebar.)
`It’s almost cliché now to refer to
`NCs and the WT asa return to the
`days of the mainframe and dumb
`terminal. It does certainly appear
`that computing has come round full
`circle, and the first impression many
`of us get is that these technologies are
`recidivist. I’m not sure that’s true.
`Yes, we’re back at the beginning in
`
`>
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page7 of 8
`
`
`
`
`
`MarketPotential Is Terrific.” VARBusiness 1411
`(May 25, 1998)
`http://www.techweb.com/se/directlink.cgi?VAR19
`980525S0051.
`
`[4] Murdock, Michelle and NC World editors
`and contributors. “The numbers are in.” NC
`World online magazine. May 1998.
`http:/www.neworldmag.com/neworld/new-05-
`1998/new-05-running.html.
`
`[5] “Network Computer Reference Profile” NC
`Reference Profile home page.
`http:/www.nc.ihost,com/nc_ref_profile.html.
`
`Reached Its Limit?” CAUSE Professional Paper
`No. 16. 1997. http://www.cause.org/information-
`resources/ir-library/abstracts/pub3016.html.
`
`[10] Petrely, Nicholas. “The network computer's
`dirty little secret.” NC World online magazine.
`March 1998.
`http://www.neworldmag.com/neworld/new-03-
`1998/new-03-straypackets.html.
`
`{11| Stern, Morgan “Terminal Compromises.”
`BYTE 23, No. 3 (March 1998), p.43.
`
`some ways. If the mainframe was
`computing’s alpha, perhaps it is also the
`omega,at least of the moment. Butis
`the thin client model of computingjust
`an economically motivated return to
`the days ofcentral control and author-
`ity? Is it merely a reversal of direc-
`tion in the pendulum that swung us,
`through the 90s, in the direction of
`distributed computing? Or is the thin
`client model instead a long-awaited
`Mark Sheehanis director of the
`liberation ofmainframe computing from
`Information Technology Center
`{6] “Well Managed Windows-Based Solutions
`the user interface constraints to which,
`Offer the Lowest TCO.” Microsoft|Market
`at Montana State University—
`Bulletin, February 1998.
`in the early 1980s, only a desktop PC
`Bozeman. He writes and speaks
`http://www.microsoft.com/windows/platform/info/
`could provide an alternative?
`frequently abouttopics in information
`gartnertco.htm.
`I guess timewill tell.
`technology management. He received
`the UMI Excellence in Writing Award
`in 1995 for his article “Pulling the
`Internet Together with Mosaic” which
`appeared in the March/April 1995
`issue of ONLINE.
`Communications to the author
`should be addressed to Mark
`Sheehan, sheehan@montana.edu;
`http:/ / www.homepage.montana.edu |
`~sheehan.
`
`REFERENCES
`{1] Gartner Group. “TCO: New Technologies,
`New Benchmarks.” Gartner Group's Managing
`Distributed Computing Research Note TCO-252
`(December 5, 1997).
`
`{2] Bell, Stephen J. “Computes Great—Less
`Filling: An Online Searcher Looks at Network
`Computers.” ONLINE 21, No. 3 (May/June
`1997): pp. 52-60.
`
`[8] Rettig, Hilary. “The Skinny on Thin
`Clients—Whether NC or Comm Device, the
`
`(7) “Desktop Clients: A Cost of Ownership
`Study.” White paper by Zona Research Inc.,
`Spring 1996.
`http://www.wyse,com/solution/tco/intro.htm.
`
`[8] “NCD Casebook: Students Sabotage Desktop
`PCs.” NCD Newsroom.
`https//www.ned.com/news/casebook/eb100397.ht
`ml.
`
`{9] McClure, Polley A., John W. Smith, and
`Toby D. Sitko. “The Crisis in Information
`Technology Support: Has Our Current Model
`
`98 http://www.onlineinc.com/onlinemag
`
`Supplied by the British Library 30 Nov 2017, 11:02 (GMT)
`
`LGEExhibit-1014/Page 8 of 8
`
`