HP NetServer LP 2000r 8-way Systems Enable Enterprise Applications - Page 10

FXVWRPHUýXVDJHýJXLGH +3ý 1HW6HUYHUý /;Uýåèíí Tier Two Servers Terminal Servers Terminal Servers Data Management Business Logic Presentation Device Management transport Tier One Clients Thin Clients transport hp-ux linux Windows 95/98 UNIX MAC OS PC Clients Windows NT/2000 (KIWTG"50"6GTOKPCN"5GTXGT"6JKP"%NKGPV A thin client is basically a simplified desktop device that provides reliable and fast access to networked resources. Thin clients have no local hard drive, no local storage and do not require a large or complicated desktop operating system to function. Thin clients rely on the speed and availability of the server infrastructure to deliver maximum computing performance to users. For example, HP Entria thin-clients offer out-of-the-box connectivity and HP's award-winning industry-standard manageability. Add to that legendary HP quality and reliability, and you have great value in a thin-client computing solution. In Figure 3 above, the central or terminal server is often referred to in industry jargon as a "Fat Server" set up to serve "Thin Clients". The server, unlike a central server in a typical ERP environment, must also provide the processing power to construct and update the graphical user interface. Thin clients address the need for an extremely reliable, secure and compact desktop device that offers the familiarity of a windows-type graphical user interface. Thin Clients are often deployed in situations where the users access one or a few specific applications. With no hard disk to fill or fail, and with application access centrally controlled, this type of computing fills many needs once the realm of "dumb" terminals or legacy PCs. The demand for thin client computers and servers to serve them is expected to develop rapidly, with growth for both devices expected to exceed 10% per year. Thin client computing has two primary server resource requirements: CPU processing power and memory capacity. The capacity of a server to support a given number of clients depends directly on both these resources to maintain adequate service levels. If either resource is exhausted, user response time will degrade severely in the case of memory exhaustion and gradually in the case of CPU. 10