Thin Client Architectures (Networking)

The thin-client computing model originated with Oracle Corp. in 1995. In this model, applications are deployed, managed, supported, and executed on a server. This allows organizations to deploy low-cost client devices on the desktop and, in the process, overcome such application deployment challenges of management, access, performance, and security.

The clients themselves come in three types: network computers, Windows terminals, and netPCs. Each type of client is suited to a particular corporate need. Since all three solutions rely on servers for their applications, realizing the benefits of the thin-client environment depends on the ability of administrators to monitor application usage, fine-tune server performance, and sometimes take control of the clients for support, diagnostics, and training.

By centralizing management of applications and using the network to deliver a GUI to desktop devices, organizations can reduce the total cost of ownership (TCO). Saving on the total cost of ownership and reducing complexity are the primary motivations for companies installing thin-client hardware. They dramatically reduce application deployment time by updating clients automatically whenever an application is installed on the server, making it easy to provide the latest applications to all users. The thin client architecture improves management and security by enabling systems administrators to control an entire network from the server. This eliminates the need for end users to install software or make changes to existing software, creating an endless array of software bugs, glitches, freezes, and crashes.

Applications of Thin Clients

Businesses that have embraced thin clients are using them for a variety of applications. Most thin clients are used to access an office suite such as Microsoft Office, but some are used to run mission-critical applications, such as accounting, transaction-processing, and order-entry applications. To a lesser extent, thin clients are also running engineering, Enterprise Resource Planning (ERP), and medical applications.

Users of thin-clients are usually task-oriented and prefer to do their work without being distracted by technology issues. These are front-line professionals, such as doctors in HMOs, accountants, engineers, and salespeople of big-ticket items, such as industrial equipment and real estate. Thin clients are also used for back-office operations supported by clerical and administrative staff, low-level salespeople, and workers on the shop floor.

Operation of Thin Clients

The operation of thin clients is fairly simple; they are dependent on servers for boot-up, applications, processing, and storage. Since some thin clients may not have a hard drive, the server provides booting service to the network computers when they are turned on. The server can be a suitably equipped PC, a RISC-based workstation, a mid-range host such as the IBM AS/400, or even a mainframe. The server typically connects to the LAN with an Ethernet or token-ring adapter and supports TCP/IP for WAN connections to the public Internet or a private intranet.

Since all applications reside on the server, they must be installed only once—not hundreds or thousands of times at individual desktops. Periodic updates to applications are conducted on the server. This ensures that every network computer uses the same version of the application every time it is accessed.

Network computers can access both Java and Windows applications on the server, as well as various terminal emulations for access to legacy data. Users accessing Java applications do so through a Java-enabled Web browser, which also gives them access to applications on the Internet or intranet. For access to Windows applications, the server must run a multiuser implementation of Windows NT, such as WinCenter from Network Computing Devices, Inc. The server’s operating system may also include terminal support for 3270, 5250, and X-Windows servers.

Role of Java

To one degree or another, Java plays a role in all of these thin-client architectures. Java was designed by Sun Microsystems to provide a cleaner, simpler language that could be processed faster and more efficiently than C or C++ on nearly any microprocessor.1 Whereas C or C++ source code is optimized for a particular model of processor, Java source code is compiled into a universal format. It writes for a virtual machine in the form of simple binary instructions. Compiled byte-code is executed by a Java run-time interpreter, performing all the usual activities of a real processor, but within a safe, virtual environment instead of a particular computer platform.

Much of the Java applications development at major corporations hinges on the Web because the Internet is increasingly being considered the foundation for network computing, providing an economical way to access corporate information from remote locations and mobile computers. A Java-enabled Web browser is the interface for accessing these applications. This arrangement minimizes the number of copies of application software and streamlines software maintenance tasks.

With the advent of thin clients and the increasing reliance on network servers, the concept of the “thin server” has emerged. This is a dedicated, special-purpose server optimized for supporting thin clients. In addition to supporting a narrow range of network applications, the thin server supports localized services to reduce network traffic congestion and provides fast access to routine applets used for many database and spreadsheet programs.

Unlike traditional server-based systems that require an investment in separate hardware and software, thin servers come complete with hardware and software at a fraction of the cost. They usually come ready to set up, install, and configure out of the box. Typically, all users need to do is plug in an Ethernet cable and set the IP address. A thin server usually comes with a browser GUI, a Java-based management application, and an embedded HyperText Transfer Protocol (HTTP)-compliant operating system.

Types of Thin Client

As noted, there are three types of thin clients: network computers, netPCs, and Windows terminals.

NETWORK COMPUTERS Network computers emphasize access to server-based applications and data via Web browsers. They combine highspeed Internet and network application performance with Java capabilities. The management software runs on the server and is used to administer the network computers and to set up their boot parameters. Some network computers can also run Windows applications via multiuser implementations of Windows NT on a PC server. Others, such as those available from IBM, can also access 3270 and 5250 terminal applications and work with applications on AIX and UNIX servers using X-Windows server support.

NetPCS Unlike network computers, netPCs are designed to run programs locally. This type of thin client comes in many different configurations. Of the three types of thin clients, the netPC is the one that most closely resembles a traditional PC. Among the first vendors to offer netPCs were Compaq and Hewlett-Packard. Both companies offer netPCs that are compliant with the DMI 2.0 specification for desktop management.

WINDOWS TERMINALS Windows terminals provide local and remote access (e.g., Internet and intranet) to Windows, Java, Web browser, and host-based applications. These machines are suited for retail, call center, financial, transaction processing, manufacturing, and office sites. In the home office, they can access a remote server using standard modems. Resident VT220, VT100 and VT52 ANSI terminal emulators make the Windows terminals ideal as character terminal replacements, ready for integration with Windows NT. Among the most prominent vendors of Windows terminals is Wyse Technologies.

Since thin clients are network-oriented, there must be a way to incorporate them into the overall management scheme. This is necessary because organizations usually have a mix of thin clients and fully configured desktop systems. Many vendors have extended and enhanced the manageability of thin clients by making them compliant with the Desktop Management Interface (DMI), specifically DMI 2.0, issued by the Desktop Management Task Force (DMTF). DMI 2.0 provides a standard way to access management data about a system, which is a key element in reducing the total cost of ownership. Version 2.0 also includes support for Intel’s Wired for Management (WfM) specification.

Wired for Management

The WfM baseline specification, developed by Intel, facilitates the central management of Intel processor-based systems over a network. WfM-enabled computers contain up to four key capabilities to support advanced systems management:

INSTRUMENTATION This is a technology that enables asset management and remote systems monitoring. It gathers information from components of the system, disk drive, and network interface, as well as I/O cards and other add-ins. The values provided by the instrumentation are described by a file called the Management Information Format (MIF) file, which is accessible through the DMI 2.0 Service Provider.

NETWORK SERVICE BOOT Also known as Preboot Execution Environment (PXE), this capability includes platform agents that allow an operating system and/or applications to be loaded on a system from a remote server. This capability is implemented as an option ROM on the NIC, or in system ROM for the LAN on motherboard products.

REMOTE WAKE-UP This is the ability to bring a system from a reduced-power state, including a full shut-down of all system functions, to a fully-powered state in which all management interfaces are available. This capability enables off-hours, remote, and automated maintenance and software upgrades. It is implemented via Magic Packet support in the NIC or LAN on motherboard (LOM) and wake-up circuitry, such as Wake-on-LAN on the system baseboard.

POWER MANAGEMENT This capability allows a PC to function in a reduced-power state, but still be “awakened” and quickly become fully operational when needed. The WfM Baseline Specification recommends that desktop systems comply with the Advanced Configuration and Power Interface (ACPI), and requires such compliance for mobile systems. ACPI compliance means that an ACPI-compliant BIOS must be resident in the system’s non-volatile memory, the platform must include an ACPI-compliant chipset, and all peripherals, such as the network interface, must support ACPI.

These capabilities are implemented via technologies that reside in the computer’s hardware, nonvolatile memory, disk drive, and network interface card (NIC).

Last Word

The promise of thin clients is that they allow organizations to more quickly realize value from the applications and data required to run their businesses, they receive the greatest return on computing investment, and they accommodate both current and future enterprise computing needs. This does not mean thin clients will replace PCs. The two are really complementary architectures that can be centrally managed. In some cases, it may even be difficult to distinguish between the two—the line of demarcation seems to be quite fluid. Overall, thin clients should provide the economies and efficiencies organizations are looking for, but these advantages will primarily come from among the user population that is task-oriented.

Next post:

Previous post: