Connecting and using thin clients. Thin clients - what are they? How to set them up

This article will discuss the technology " thin clients", describes the benefits of their use, types of clients and examples of their use.

On a typical network small organization There are usually one and a half to two dozen computers, one to two servers and a small number of other devices. Such a system is maintained by one or two system administrators. As long as the scale of the organization does not change significantly, this is quite sufficient.

But what happens when the number of network elements increases significantly (at least to 50-70 computers)? The number of computers grows, the load on the server and, in particular, on the data storage system increases, and the network begins to “slow down”. To increase productivity, new server equipment and new computers are purchased. Naturally, system administrators are hired, because one or two people are in principle unable to maintain such a quantity of equipment. Moreover, the expansion of the staff should be very significant, since, as any IT specialist knows, the most mysterious phenomena are constantly happening to users’ computers. To this must be added the cost of the purchased equipment itself, since modern software either refuses to work on computers older than two to three years, or works, but at an unsatisfactory speed.

What is most interesting is that all these time-consuming and expensive measures do not bring the desired result - the network is working slower and slower, the number of failures is constantly increasing. What is the reason? The reason is the wrong principle of organizing a corporate network.
Despite its high popularity in the West, terminal networks are still little known in Russia. The main reason here is rather psychological.

The “thin client” itself (hereinafter referred to as the terminal) is a simple device designed to work in an SBC (Server Based Computing) environment. During operation, they interact with applications deployed on the server through terminal emulation software that displays the information transmitted by the server. Technically, these are compact (about the size of an average book) computers that are not very tall. technical parameters(approx. 500MHz, 128 RAM). There are no terminals disk drives and storage devices. Thus, such computers are not capable of operating without powerful server hardware.
This is precisely the reason for the low popularity of terminal solutions - good server They are not cheap and in the short term the terminals do not look like an attractive solution compared to traditional computers (“thick clients”).
The situation changes radically if you conduct a small analysis of the costs of network infrastructure over a significant period of time. A fairly typical breakdown by expense item looks approximately like this:

• purchase of equipment and software – 13%
• installation – 1%
• maintenance – 3%
• modernization of basic software – 3%
• application software modernization - 11%
• informal administration – 14%
• network administration – 55%

It is not difficult to notice that the main funds are spent not on the purchase of hardware, but on its proper configuration and support in working condition. And that is why terminal solutions benefit greatly from the parameter. Using terminal access, the administrator no longer has to run around the entire organization and try to unite a conglomerate of computers with different configurations, settings and software into a single workable system. The process of installing, configuring and integrating the next terminal takes literally a few minutes, without leaving your workplace (as a rule, standard configuration terminals are used within the same organization and the entire setup involves creating an account on the server side).

Many people believe: “Since all calculations are performed on the server side, then its performance should be equal to the total performance of all computers that users previously used.” But this is not true - we can confidently say that 95% of the time personal computer is used at 5%, having a pronounced peak load pattern. Moreover, these peaks from all clients are not simultaneous. In addition, if performance is an issue at all, it is much more efficient (and cheaper) to increase server resources by 50% instead of increasing the resources of fifty clients by 20% each.

The following main advantages of “thin clients” can be identified:

Savings, investment protection
The terminals do not require modernization; the terminals do not contain most of the expensive components - hard drives, large amounts of memory, external video cards, etc. The total cost of system ownership is reduced by reducing the time required to service user workstations, capabilities quick recovery failed workplace, energy savings (up to 80%)

Reliability

Longer time between failures. The absence of mechanical components, as well as the simplified architecture itself, increases the reliability of the system. Eliminates the possibility of information loss due to station failures (all information is stored on the server)

Long service life

Terminal stations are significantly less susceptible to obsolescence than conventional PCs

Security of information storage

High level of system security. The absence of disks and drives significantly reduces the risk of information leakage and viruses being introduced into the system.
There is no data transmission over the network; only the screen image is transmitted to client sites. The possibility of software encryption of data without the use of additional equipment eliminates the possibility of unauthorized interception;
Centralized storage of data and settings simplifies procedures backup There is no need to worry about the safety of data and programs on workstations.

Ease of administration

Simplify administration and reduce user maintenance costs. Users cannot influence the stability of the software at their workplace. Administration of the terminal system is completely centralized. To resolve user problems, the administrator just needs to connect to the user session. The control and management of the software used in the company is simplified. Simple organization of user control and restriction of unwanted activities.

Remote access capability

The user accesses his virtual desktop from any terminal connected to the server. You can connect a thin client even from your home; just connect it to a terminal server (for example, via the Internet). Preliminary and one-time setup takes only a few minutes, after which the user immediately gets to his workplace with already installed programs (on the server).

Freeing up resources, reducing network load

The load on the local network is reduced, since only screen states are transferred to the terminal, while significant amounts of data can be transferred to the personal computer. In case of a lack of computing resources, it is enough to upgrade the terminal server, and not the entire fleet of personal computers.

Ergonomics

The terminals operate silently, since thin clients, as a rule, either have no fan at all or are equipped with only one fan. Small size and ergonomics. Thin clients have this name for a reason. They are usually no larger than a large book and do not take up much desk space.
There are three types of terminal solutions on the market today:

X-terminals

X terminals are similar to the old non-intelligent displays that were once widely used as access devices to minicomputers and mainframes. The X-terminal uses the X-Window protocol to interact with those operating on Linux server or Unix applications. It runs the X server program and uses local fonts to display information. This operation requires more CPU resources and a larger amount of RAM than normal operation of thin clients of other categories. In addition, X terminals must maintain state information about their work sessions.

Windows terminals

Windows terminals run one or another version of the Windows OS and support the ICA and RDP protocols. They load their OS from ROM or from a server (where its image is stored) and visualize the screens of applications running on the server. Windows terminals use thin client software - client programs for Microsoft Terminal Services and Citrix. Although rendering application screens on a Windows terminal requires less CPU and RAM resources than displaying them on an X terminal, these resources still need to be available in a fairly large amount.

Sun Ray thin clients are designed to work in Solaris environments. Unlike X or Windows terminals, they do not store information about the state of their sessions (it is stored on the server). The Sun Ray product operates under firmware control that communicates with Sun Ray servers. In addition, these “thin” clients work with smart cards. These cards are used for user authentication, and can also support certain applications and store data. Running on the Solaris server and managing user sessions, Session Manager (part of the Sun Ray server software) sends appropriate video information to Sun Ray devices (see figure). Because the clients in question do not store state information about user sessions, sessions can be initiated or resumed using any client. This way the session is able to sort of follow its user

Network requirements

During typical operation, traffic from client to server does not exceed one kilobyte per second, the maximum value recorded in the test session is 1006 bytes/s. Traffic in the opposite direction (server-client) is several tens of kilobytes per second. Maximum value achieved during the session - 106664 bytes/s (achieved when opening an IE window with graphics and dynamic flash banners on mail.ru). The average traffic value is about 5-6 KB/s (working Internet Explorer, viewing WinWord documents without graphics, opening and working programs with standard elements user interface). Such low traffic is achieved not only by compression of the transmitted data (up to 300%), but also, mainly, by the fact that during the session only commands are sent to the client to locally display user interface elements (windows, buttons, font design) instead of their image. Exceeding the maximum channel bandwidth does not cause a crash, but only slows down client screen updates. This allows, if necessary, to work even through a modem connection with a bandwidth of 2-5Kbit/s. If we take the nominal operating bandwidth of an Ethernet network as 100 Mbit, leaving the necessary margin of safety for critical traffic of approximately 2-3 MB/s, then this bandwidth makes it possible for either 20-30 clients to work in the toughest mode without the slightest delay in screen updating, or up to 500 clients in case of regular office work without active dynamic graphics requiring constant transfer graphic images to the screen. Taking into account the fact that even in the case of dynamic graphics, the channel load has a peak nature, it is quite possible that these values ​​will be slightly exceeded without compromising the convenience of the clients (the screen loading peaks of some machines will occur during periods of waiting for the user to respond to other clients).

Application

Thin clients can be used wherever a large number of users perform the same type of office or specialized tasks that do not require large PC resources. These could be, for example:

• operator hall
• office terminals
• classrooms
• also internet cafes and so on.

The use of terminals is impossible if the work involves processing significant amounts of data - working with graphics, sound, video, performing calculations, etc. Applications that generate excessive traffic (watching videos, modern 3D games) are also not applicable.

Conclusion

Thus, the advantages of thin clients make them quite attractive to many organizations. You just need to clearly define for yourself the advantages and limitations of the terminal approach to organizing workplaces. It is also important to note that the total cost of ownership (TCO - Total Cost of Ownership) turns out to be significantly lower (according to the Gartner Group - by 5-40 percent) when using thin clients in workplaces, rather than full-fledged computers. After all, TCO consists not only of the costs of purchasing equipment, but also of the costs of administering and upgrading this equipment. Reducing the likelihood of equipment failures also leads to a decrease in TCO.

The concept of thin terminal clients is not new. Indeed, why equip the user’s workplace with relatively productive hardware, purchase a license for a client OS, install application software, antivirus, and ensure the proper level of protection? workstation and data, if the user performs all his operations on the terminal server, essentially without using local resources (except for peripheral devices). In this article we will conduct brief overview domestic solution for organizing thin terminal clients - WTware.

WTware is an optimized Linux-based distribution that includes all the necessary drivers and clients for connecting to Windows terminal servers (rdesktop), Linux (xrdp), Hyper-V VDI, Mac Terminal Server.

Main advantages of WTware:

• Low hardware requirements. WTware can be run on almost any computer with at least 48 MB RAM (64 MB required for optimal performance). For Raspberry Pi 2 exists free version WTware (http://winterminal.com/ru/)
• It is not necessary to run the client hard drive. Both network downloading and downloading from any media are supported
• Easy to install and configure the client part; does not require the administrator to have knowledge of Linux administration
• Centralized terminal configuration management
• Supports a wide range of equipment. Possibility of forwarding local printers, barcode scanners and other peripherals to the terminal session
• Support for remote connection to the terminal console by technical support services (via VNC)
• WTware is a Russian product, which means that all documentation and technical support are also provided in Russian.
• Possibility of simultaneous connection to 4 terminal servers (switching between sessions using combinations Win+1 – Win+ 4)

Let's consider the “quick” start procedure for using the WTware solution to organize a workplace with a thin terminal client in a typical office network.

WTware client download options

Before you begin setting up and deploying WTware, you need to select your preferred method for booting thin clients. WTware can boot from almost anything, be it:

• Hard drive
• CD-Rom
• Flash drive
• Diskette
• Network card with BootROM

In most cases, it is preferable to use network boot, because... this makes deployment and centralized client management much easier. This is the loading option we will consider.

Note. In the event that you need to connect single clients from remote offices connected via slow communication channels to the terminal, you can use downloading from physical media. If such offices have several clients and any server, it is still worth considering deploying your own WTware TFTP server on it.

We also note that the manufacturer’s website indicates the ability to load terminals via HTTP, which should reduce the load on TFTP with a large number of clients (more than 300) and improve loading on slow and unreliable communication channels.

Boot processWTware

To run the WTware client on the user’s computer, you need to:

• Download distribution binary files from the server (via TFTP) or local media
• Get network settings from a DHCP server or from local configuration files
• Receive a configuration file from the server (via TFTP) or load it from disk

Installing the WTware server part

Let's start by installing the server part of the WTware system. In our case, we decided to install it on a DHCP server running OS.

Download the distribution kit from the developer’s website - at the time of writing, the version is wtware.5.4.8.ru.exe (226 MB) and start the installation.

We indicate the path to install the configuration files (by default, C:\ProgramData\WTware) and the program itself (C:\Program Files (x86)\WTware).

• Service WTFTP– necessary for downloading over the network, keeps a log of calls and allows you to diagnose problems
• Service WTUSBIP– WTware USBIP Initiator service is used for automatic connection USB devices terminal
• Services WTDHCP– assigns IP addresses to terminals, required for downloading over the network

Because We will use our own existing DHCP server, so we will not install the WTDHCP service. Setting up an MS DHCP server is described in the section.

Advice. If your network has not yet deployed a DHCP server, it makes sense to use the built-in DHCP servers of WTware (WTDHCP). Using WTDHCP allows you to quickly deploy and launch a DHCP service for small network. The WTDHCP service is configured during installation and subsequently using a graphical utility - the WTware configurator (win32.exe), the capabilities of which are discussed.

Note. If your network is divided into segments, each of which will contain thin clients, there is no need to install its own DHCP server in each. One server can serve a large number of zones (subnets). Forwarding DHCP packets between segments is possible via .

Let's start the installation.

After installing WTWare, two additional services will appear on the system:

• WTwareTFTP– executable file C:\Program Files (x86)\WTware\Bin\wtftp.exe – uses local port UDP/69
• WTware USBIP Initiator— C:\Program Files (x86)\WTware\Bin\wtusbip.exe – TCP/780 port

Configuring DHCP server parameters

It is assumed that our network has already deployed and used a DHCP server on any server Windows edition. We launch the DHCP management console (dhcpmgmt.msc) and find the DHCP zone we are interested in (in our case, the zone name is Managers). We need to specify additional zone settings necessary for network booting of diskless terminals.

In the zone settings, you need to additionally specify two parameters:

• 066 (Boot Server Host Name) – here you specify the IP address of the server on which the WTware TFTP server will work (for us it coincides with the address of the DHCP server)
• 067 (Bootfile Name) – here is the file from which the terminal should start loading. To boot using PXE (if BootROM is built into your network card or motherboard by the manufacturer) we set the parameter value 5.4.8/wtware.pxe. The wtware.pxe file is located in the 5.4.8 subdirectory of the tftp server root (by default, the tftp root is located in the C:\Program Files (x86)\WTware\TFTPDROOT\ directory)

Note. For the Etherboot bootloader (when using the BootROM emulator), you need to specify another file as the value of option 067 - 5.4.8/wtshell.nbi

After configuring these two parameters, the DHCP server provides the client with all the necessary information to download over the network.

Note. In the event that the WTWare terminal will be installed on a flash drive or local disk, you can prohibit users from changing the configuration of their station. To do this, you need to password protect the WTware Setup menu. To do this, you can use another DHCP zone option - 018 (Extensions Path). This field indicates the password hash obtained using a special utility. The purpose of specifying the hash is to prohibit transmission of the password in clear text in a DHCP response.

Configuring WTWare terminal parameters

Configuring WTWare terminal clients that use network boot is done using configuration files. The client configuration is formed from three files:

• System-wide configuration file all.wtc(C:\Program Files (x86)\WTware\TFTPDROOT\Everyone)
• Personal configuration file config.wtc(stored in the personal directory of each client, identified by MAC address, for example (C:\Program Files (x86)\WTware\TFTPDROOT\Terminals\00.50.56.BB.AD.80)
• Included files defined in the file list.wtc

In the all.wtc file you need to specify parameters that are the same for all terminals.

For example, you can specify the addresses of terminal servers available for connection by specifying their IP address

server=10.24.181.44

or DNS name server (provided that clients obtain the address of the name server on the network via DHCP)

server= msk-term-1c.site

Or you can allow the user to specify the name of the terminal server to which to connect.

Note. By default, the RDP client is launched on the client, but it is possible to launch a browser on the thin client Google Chrome. In this case, the client must have at least 512 MB of RAM, and the following lines are specified in the configuration file (we will also set the proxy server address for the browser):

application=chrome
chrome_proxy=192.168.1.23:3128

So that on the terminal server instead of the desktop it will immediately open specific application, you need to specify the shell parameter in the configuration file:

For example, to launch the Directum client you need to specify:

shell = C:\Program Files (x86)\DIRECTUM Company\DIRECTUM 5.1\SBRte.exe -S=msk-drc01 -D=DIRECTUMDB

Each client's individual configuration files are stored in the C:\Program Files (x86)\WTware\TFTPDROOT\Terminals\ directory. A personal directory with its MAC address is created for each client. It is in this directory that the client will look for the config.wtc file with its configuration.

The developer's website presents more than these and other configuration file parameters.

Setting up and working with the WTWare client

So, the configuration of the server part is completed, let's move on to setting up the client. In the BIOS/ of the computer that will be used as a thin client in the section in which the search order is configured boot devices, indicate the highest priority to network boot with PXE (Network boot, LAN boot).

Save the changes and reboot the system. If everything is configured correctly on the WTware and DHCP server side, the client must obtain an IP address from the DHCP server and, according to the parameters we have configured, perform a network boot from the specified .

When you start it for the first time, you can configure the terminal (F10 – terminal setup wizard).

We will be asked to select the video card driver and other display options. Preferred settings can be saved to the client’s personal configuration file on the server. In this case, next time you will not need to manually edit the display settings.

To do this, on the TFTP server in the directory C:\Program Files (x86)\WTware\TFTPDROOT\Terminals\00.50.56.BB.AD.80 (a directory with a name containing the client’s MAC address), we will create a file config.wtc, in which there will be client settings are specified:

video=VESA(F)
bpp= 16
display = 800x600

The next time you boot, the terminal will automatically boot with these parameters.

If the terminal server address is specified in the configuration file, the WTWare client automatically initiates an RDP connection. All that remains is to log in to the server and its desktop will open in front of us.

If the configuration file provides the ability to independently select a terminal server, the client can specify it manually.

To diagnose client operation, a small web server operates on each terminal. To open the diagnostic page, just type the client’s IP address in the browser. On the web page that opens, you can see the current client settings, the state of its components, logs, shutdown/restart buttons for the client, etc.

By default, access to this page is not limited. To allow connections to the web server only from certain addresses, you need to specify the line in the configuration file:

httpd = 10.10.1.55, 10.10.1.56

WTware graphical configurator

In addition to managing through text configuration files, it is possible to manage the settings of the system and client terminals from a separate graphical application– WTware configurator (C:\Program Files (x86)\WTware\Bin\win32.exe), which allows you to more conveniently work with text config files.

What can you use this utility for:

In the vast majority of cases, using the configurator is preferable to manually editing configuration files, because Navigation through the structure of configuration files is simplified and the likelihood of errors is reduced.

WTWare Licensing and Pricing

WTWare licenses are tied to the MAC address of the computer's network card. All licenses must be written to the server in the wtware.lic file.

The cost of a WTWare license for one workplace depends on the number of clients and starts from 1000 rubles (for the number of clients from 1 to 9) and ends at 350 rubles (for purchasing more than 100 licenses).

Conclusions

WTware leaves the impression of a high-quality and solid product that allows you to deploy thin terminal clients without significant costs. The solution from WTware impresses with its simplicity and simultaneous flexibility in terms of centralized administration and deployment. And the low cost of licenses almost immediately leaves all competitors behind.

Of the free analogues of WTWare for organizing a thin client, you can recall Thinstation, but the latter is significantly inferior in manageability and is much more difficult to deploy.

Thin client called a device for inputting and displaying information (terminal). Physically, a thin client is compact and silent computer without hard drive, the main operating system of which is loaded on the server. All user applications run on the terminal server (application server), but this is completely transparent to the user. Since the entire computing load falls on the server, the thin client has a minimal hardware configuration without any loss of performance.

What are thin clients used for?
Thin clients are used in organizations where most users use computers to perform the same type of tasks: working with databases, information catalogs (shops, pharmacies, libraries), working as bank terminals, etc.

. Terminal servers use system-wide software Windows, Linux, Solaris. Thin clients operate under control operating systems Windows CE, Windows XP Embedded, Linux. The thin client software is located directly in the thin client in the built-in flash memory. In models entry level, which do not have built-in memory, the software is downloaded from the server when the thin client is turned on ( this technology implemented in Aquarius thin clients). The latter option, however, is impractical if the communication channel between the server and the thin client has low bandwidth, or is charged based on the volume of transmitted traffic.

What operating system is on the terminal?
The terminal operating system is “firmware” in a small disk-on-module device (flash memory 64MB-1GB). It provides the basic functionality of the client: initial boot, correct operation of the video adapter, audio, operation of peripheral devices connected directly to the terminal client (mouse, keyboard, local printers, USB flash drives). Also, the thin client operating system may contain an Internet browser that can operate autonomously (without a terminal server). When switching to terminal mode, the client begins to work with the server operating system, an individual session of which is launched on the terminal server. From this point on, the terminal becomes simply a means of displaying and entering information.

What software licenses are needed?
To organize the operation of a group of terminals with Microsoft software, in general, you will need the following licenses:
Licenses for embedded OS on terminals (Win CE 5.0 or Win XP Embedded), license for server OS (Windows Server 2008), client access licenses (Windows Server CAL 2008) - the required number of licenses is equal to the number of terminals, terminal access licenses (Windows Trmnl Svcs CAL 2008) - the required number of licenses is equal to the number of terminals or users. Licensing of application programs, as a rule, is carried out according to the principle of as many users (terminals), as many licenses.

Pros and cons of the terminal method of building a network(in case of using Windows OS).
Pros.

• reduction in initial acquisition costs due to minimum requirements to configuration;
• significant reduction in power consumption - a typical thin client has a power consumption of only 10W (versus 250-350W for a PC)
• unification - all clients have the same set of software;
• ease of implementation of tasks - there is no need to configure each computer separately, since clients are centrally managed. The system administrator performs all settings for managing thin clients centrally on the server;
• saving time for a system administrator who maintains absolutely identical computers, the likelihood of breakdowns of which is minimized, and all programs are installed on the server;
• scalability - a system image created once for the entire group of users to work with allows you to maintain an easily scalable network. You can install as many PCs as you need, while adding new workstations requires minimal effort;
• security and fault tolerance. When the terminal boots, it receives an operating system “from the manufacturer”, which is configured only by the information support department. All user information is stored on the server on a RAID array and is regularly backed up, which increases fault tolerance;
• protection against information leaks - no local storage - no ability to make copies of documents on removable media information (unless otherwise permitted by the system administrator).
• any terminal is an analogue of a powerful workstation; all programs are executed locally on a fast terminal server.
• ease of increasing computing power - there is no need to upgrade the terminal. Since it is only a device for input and display of information, without processing anything itself. If there is a lack of computing resources, it is enough to upgrade the server (usually this is more profitable than upgrading N full-fledged workstations), and the new resources will be available to all terminals at once.
• the ability to access your virtual desktop and all documents from any terminal connected to the server. Since all information is stored on the server, it is enough to authenticate in the system (enter your username and password) from any terminal.
• no problems during power outages. Since all information is stored on the server, it is enough to equip it with uninterruptible power supply devices. Loss of energy in the workplace will only lead to a temporary inability to see what is happening on the terminal screen. Here we can give the following analogy - after all, when we turn off the monitor, nothing happens to us open source software? After power is applied (or when restart terminal) the user will return to that state running programs, which remained at the moment the terminal was turned off.
• acceleration of some programs that place increased demands on network bandwidth. Good examples of such programs are 1C Accounting and Parus. When its server and client parts are located on the same machine, the bottleneck is eliminated - sending data over the network during client requests to the database, and programs begin to work much faster.
• quick deployment of a new workstation - you can connect a thin client even from your home, just connect it to a terminal server (for example, via the Internet). Preliminary and one-time setup takes only a few minutes of time, after which we immediately find ourselves at our workplace, with programs already installed (on the server).
• quiet operation - usually terminals do not include mechanical components, such as hard drives and fans (cooling is carried out passively), so they do not make any noise
• longer time between failures . The absence of mechanical components, as well as the simplified architecture itself, increases the reliability of the system as a whole, which is important, given the much longer service life of terminals compared to workstations.
• small size and ergonomics. They are usually no larger than a large book and do not take up much desk space.
• at work you have to work. - It will be impossible to play 3D games or watch videos. Firstly, they will not be on the server and it will be impossible to install them yourself (due to restrictions set by the administrator on the installation of additional software). Secondly, the network bandwidth is not enough for acceptable screen refresh rates for these applications.
• a complete terminal access system is not tied to any specific brand of equipment and consists of thin clients (terminals) - compact devices installed at workplaces, a terminal server for running user applications, and, finally, software that is already built into the server operating system a system like Microsoft Windows Server or Linux.

• A thin client is not a panacea for everything.
  The terminals are not designed to perform heavy tasks involving complex calculations (for example, AutoCAD and other modeling systems) or generating high traffic for transmission to the client (for example, watching videos). In the first case, this is due to the heavy load on the server’s computing power (it can serve very few clients), in the second, due to network bandwidth. In this case, you need to use full-fledged workstations. By the way, modern 3D games fall into both categories.
• You will still have to pay.
  The lower cost of the terminal is compensated by the high price of the server. After all, this machine must be powerful enough to perform the tasks of many thin clients connected to it. To be fair, I note that the dependence of server power on the number of working clients is not linear. Most typical tasks (for example, several copies of MS Office in memory) use the libraries of the already running first copy for their work, so the RAM requirements will be relatively low.
• Server OS - MS Windows.
  With all the ensuing consequences in the form of considerable demands on server performance only for the OS’s own needs. But it can be scaled by distributing the client load across several servers in the case of MS Windows Advanced Server or Data Center.
• In general, everything runs on one server computer.
  Therefore, all possible measures must be taken to ensure its trouble-free operation and data safety.
• The need for a constant communication channel
  In some cases, it is not necessary for a workstation to have a permanent, much less a fast, communication channel. The terminal requires constant communication with the server. On average, you need a channel with a throughput of at least 20 Kbps.

Law of the cycle:
History develops in a spiral.

This statement from general philosophy, familiar to us from school, also applies to computer technology. Let's delve into history. In the beginning there were mechanical abacus with dominoes, then mechanical computers appeared. Later appeared vacuum tubes and the first large computers based on them, which occupied entire buildings. Then the transistor was invented, may its name be glorified. Let me note that the word “transistor” in those days meant more than just one of the many cells on a silicon wafer, each of which was no larger than 0.18 microns in size, and millions of them could be placed on a wafer with an area of ​​​​a square centimeter. No, the transistor was a solid radio component in a separate housing with dimensions of at least a centimeter. But it was still a breakthrough. As a result, relatively compact computers appeared, occupying “just” a few cabinets and possessing crazy speed for those times. In addition to their enormous size, they had another important drawback: price. Therefore, such computers were not available to anyone who wanted them. We had to use computers together. First in time sharing mode, one by one. And when keyboard input and displays were invented, as well as multitasking operating systems (OS), it became possible for many users to work simultaneously on the same machine. That is, at the workplace there was only a terminal, a keyboard for entering data and a display for displaying it, and all information processing took place on a large computer (mainframe).

But science did not stand still, microcircuits became smaller and smaller, and the cost of their production decreased. As a result of this rapid progress, we have come to inexpensive and powerful personal computers that are owned by almost everyone. Relatively inexpensive. After all, if we add up the costs of an organization that needs 20, 50 or more machines, add the cost of the software installed on them, plus the costs of the people servicing these computers - and we will get a very large figure.

It’s not for nothing that I mentioned that modern personal computers (PCs) are very powerful. After all, for most office and corporate tasks (such as MS Office, 1C Accounting, Parus, using Internet Explorer for WEB surfing, email, etc.) high power is not needed. But looking at the PriceList of any reputable computer company, we are unlikely to be able to find a typical computer configuration with a processor frequency below a gigahertz. But on average (in typical office tasks) it will be loaded at only 10–15 percent. Moreover, sometimes the bottleneck (with such modern and fast computers) is network bandwidth. A good example of this is 1C Accounting clients generating large network traffic when accessing the database while working.

In addition, no matter how modern a computer purchased today may seem, after 2-3 years it can hardly handle new software. Therefore, an outdated computer has to be upgraded or, simply put, replaced with a new one.

This is why thin clients have become increasingly popular over the past few years. What is a thin client?

In a nutshell, this is a stand-alone diskless device with a power source to which input devices (keyboard, mouse, smart card reader, etc.) and information output devices (monitor, printer, speakers, etc.) are connected. The device is designed for input information and sending it to the server, as well as to display information received from the server. The terminal is connected to the server via an Ethernet local network, via a modem or via serial communication lines. The server must have a multitasking and multi-user OS installed, and the server must be powerful enough (compared to the terminals connected to it). In general, the thin client itself does not process information; all calculations occur on a remote server. That is why the power requirements for such devices are low. In principle, even 286 and 386 computers can serve as a thin client, on which you can install DOS or Windows 3.11, as well as the program itself for connecting to the server. As a result, we get a fast machine running Windows 2000 (if this operating system is installed on the server) for working in typical office tasks.

Now let's draw parallels. A network of autonomous terminal devices connected to a powerful server computer. Terminals only receive data from input devices and send them to the server, and also display the data received from the server on the monitor. All information is processed on a remote machine terminal server. Yes, yes. I remember exactly computing centers 70s with mainframes. Of course, these are no longer monochrome displays operating in text mode. There is not only a keyboard on the table, there is sound. And the range of tasks performed has expanded. But the idea remains the same.

Of course, the thin client market is not limited to the MS Windows operating system family. But thanks to the prevalence of products such as MS Office and of course MS Windows with IE, the ubiquitous Windows is installed in many organizations. That is why many are interested in terminal solutions based on this OS.

There are two protocols for interaction between thin clients and terminal servers on the market. The first, ICA (Independent Computing Architecture) from Citrix, appeared on the market a little earlier. The second, RDP (Remote Desktop Protocol), was developed by Microsoft and ships with WindowsNT TSE, Windows 2000 Server and higher. ICA is considered more universal, and its support is not limited to win32 versions of Windows operating systems. It also supports different versions*nix OS, Macintosh and even Java clients. As for the rest, by now the functionality of both protocols is almost equal. Although there is an opinion that Citrix provides greater freedom and convenience in administering a terminal network.

I will try to summarize the pros and cons of the terminal method of building a network (in the case of using Windows OS).

• Any terminal is an analogue of a powerful workstation.
  After all, all programs are executed locally on a fast terminal server.
• Saving money both on the cost of the thin client itself and on the cost of software.
  Indeed, the cost of the terminal is somewhat lower than the average workstation. The software is installed only on the server computer, although all terminal owners (or some, depending on access rights) can use it.
• Ease of increasing computing power.
  There is no need to upgrade the terminal. Since it is only a device for input and display of information, without processing anything itself. If there is a lack of computing resources, it is enough to upgrade the server (usually this is more profitable than upgrading N full-fledged workstations), and the new resources will be available to all terminals at once.
• Reduction of workload for the network administrator.
  After all, you will only have to administer one server, and not N workstations scattered throughout the organization’s office(s).
• The ability to access your virtual desktop and all documents from any terminal connected to the server.
  Since all information is stored on the server, it is enough to authenticate in the system (enter your username and password) from any terminal.
• No problems during power outages.
  Since all information is stored on the server, it is enough to equip it with uninterruptible power supply devices. Loss of energy in the workplace will only lead to a temporary inability to see what is happening on the terminal screen. Here we can give the following analogy: after all, when we turn off the monitor, nothing happens to open programs? After power is supplied (or when the terminal is turned on again), the user will return to the state of running programs that remained when the terminal was turned off.
• Acceleration of some programs that place increased demands on network bandwidth.
  Good examples of such programs are 1C Accounting and Parus. When its server and client parts are located on the same machine, the bottleneck sending data over the network during client requests to the database is eliminated, and programs begin to work much faster.
• Good security implementation.
  Again, all files are located on one server, there is no information on thin clients, access rights are differentiated using system tools. The user cannot transfer information to a floppy disk, since there are usually no disk drives or access to them is limited. That is, only the server itself needs to be protected, and physically it and the terminals may be located geographically far from each other, for example in neighboring buildings.
• Quick deployment of a new workplace
  You can connect a thin client even from your home; just connect it to a terminal server (for example, via the Internet). Preliminary and one-time setup takes only a few minutes of time, after which we immediately find ourselves at our workplace, with programs already installed (on the server).
• Quiet operation
  Typically, terminals do not contain mechanical components such as hard drives and fans (cooling is passive), so they do not produce any noise.
• Low power consumption of the terminal
  The architectures used have low power consumption. For one computer this may not be noticeable. What if there are 100 of them?
• Longer time between failures.
  The absence of mechanical components, as well as the simplified architecture itself, increases the reliability of the system as a whole, which is important, given the much longer service life of terminals compared to workstations.
• Small size and ergonomics
  Thin clients have this name for a reason. They are usually no larger than a large book and do not take up much desk space.
• You have to work at work.
  You will not be able to play 3D games or watch videos. Firstly, they will not be on the server and it will be impossible to install them yourself (due to restrictions set by the administrator on the installation of additional software). Secondly, the network bandwidth is not enough for acceptable screen refresh rates for these applications.

Cons. Reverse side medals are also present.

• The thin client is not a panacea for everything.
  The terminals are not designed to perform heavy tasks involving complex calculations (for example, AutoCAD and other modeling systems) or generating large traffic for transmission to the client (for example, watching videos). In the first case, this is due to a large load on the server’s computing power (it can serve very few clients), in the second, due to network bandwidth. In this case, you need to use full-fledged workstations. By the way, modern 3D games fall into both categories.
• You will still have to pay.
  The lower cost of the terminal is compensated by the high price of the server. After all, this machine must be powerful enough to perform the tasks of many thin clients connected to it. To be fair, I note that the dependence of server power on the number of working clients is not linear. Most typical tasks (for example, several copies of MS Office in memory) use the libraries of the already running first copy for their work, so the RAM requirements will be relatively low.
• Server OS MS Windows.
  With all the ensuing consequences in the form of considerable demands on server performance only for the OS’s own needs. But it can be scaled by distributing the client load across several servers in the case of MS Windows Advanced Server or Data Center.
• In general, everything runs on one server computer.
  Therefore, all possible measures must be taken to ensure its trouble-free operation and data safety.
• The need for a constant communication channel
  In some cases, it is not necessary for a workstation to have a permanent, much less a fast, communication channel. The terminal requires constant communication with the server. On average, you need a channel with a throughput of at least 20 Kbps.

To summarize, we can say that the advantages of thin clients are attractive enough for many organizations to use them. You just need to clearly define for yourself the pros and cons of the terminal approach. It is also important to note that although the low cost of thin clients can be offset by the high cost of the server part, the total cost of ownership (TCO - Total Cost of Ownership) turns out to be significantly lower (according to the Gartner Group by 5-40 percent) when used in workplaces. thin clients, not full-fledged computers. After all, TCO consists not only of the costs of purchasing equipment, but also of the administration of this equipment and the costs of modernization. This also includes losses due to equipment failures.

As a striking example of a terminal solution, consider a thin client from a Russian company.

The photo shows the appearance (shot using a blue filter) of the company's AK Windows GP terminal. Please note that this is a Russian company. The price of such a thin client is only about 300 US dollars (at the time of publication), which is significantly lower than its Western counterparts. People familiar with the price level for old components may object: but when assembling a Pentium-class computer ][ you will have to spend only $200, and a Pentium actually costs a penny! Yes, but do not forget that if we build our own, we will get a personal computer operating in terminal mode. With larger housing sizes, power consumption, hard drive, coolers and resulting noise, old components, etc. And with the resulting reliability. And thin clients were initially developed specifically for their specific range of tasks, do not have mechanical parts and are potentially more reliable. Don’t forget about technical support we live in Russia the company is nearby.

Specifications of AK terminal GP.

The attentive reader has already noticed that the processor frequency of the GP thin client is only 300 MHz. In fact, nothing more is needed, because the terminal is practically not busy processing information. However, the company also offers models with higher processor frequencies, based on the VIA C3 processor.

A look from the inside

The back panel of the terminal with interfaces looks like this:

And inside the case there is only one motherboard, on which the entire element base devices.

Instead of a hard drive, a 16 MB flash drive is used. This is where the Windows CE OS is initially loaded. It is its windows that are visible on the screen before connecting to the terminal server.

The heart of the system is the GX1 processor from National Semiconductor. This is the company's first chip created using 0.18 micron technology and has very low heat dissipation of 0.8-1.4 W, depending on the operating mode. The processor was originally designed for installation in various terminal systems, Internet set-top boxes, etc.

The processor core is built on the x86 architecture, it includes fixed-point and floating-point arithmetic units and has a 16 KB first-level cache. The processor integrates a VGA controller with a 2D graphics accelerator and additional buffers, interfaces to external controllers for MPEG1 video playback. The controller is also integrated into the processor SDRAM memory and PCI buses. The chip also supports the MMX instruction set and can operate at frequencies from 200 to 333 MHz. The processor has built-in basic support for audio functions (some of them are handled by an external controller). Due to low heat generation, only passive cooling of the processor (radiator) is used.

The role of the south bridge is performed by the CS5530A chip (PCI-to-ISA bridge). It also integrates two IDE controllers and a USB bus for two ports. Additionally, the CS5530A connects to the GX1 processor's video controller interface for additional video acceleration and output to analog or digital displays. The CS5530A also has a built-in interface for connecting an external AC"97 audio codec. The chip has low power consumption and low heat generation.

To support other I/O functions, a PC97317 controller from the same manufacturer is used, connected to south bridge. The PC87317 is a single-chip controller for most standard ISA/EISA/MicroChannel peripherals. The chip contains an advanced Real-Time Clock module, a floppy disk controller (FDC controller), a keyboard and mouse controller, two serial port(UARTs) with support for infrared communication interface (IR), an IEEE 1284 parallel port controller and several other controllers.

The terminal itself is assembled in a small case and, as can be seen in the photo of the board, does not have mechanical moving elements such as fans and hard drives. Therefore, we can safely talk about zero noise level when operating this terminal.

Preparing for the first launch

MS Windows NT TSE or Windows 2000 server (and higher) must be installed on the server. Additionally, you should install the terminal services service in the installation panel under the removal of OS components. After this, you just need to create the required number of user accounts.

Setting up the thin client itself is very simple.

When you launch the terminal for the first time, it will display setup wizard prompts. With its help, a fixed or dynamic IP address of the client is set.

Optionally, the addresses of DNS and WINS servers and the presence of a local printer (connected directly to the terminal) are specified.

The same wizard allows you to set the resolution and refresh rate of the monitor screen. After this, the wizard proceeds to configuring (specifying) the remote terminal server and its parameters. Moreover, nothing prevents you from creating several records to different servers and then switching between them.

First, you should select the connection protocol type: RDP (native protocol for WinNT TSE/Win2K Server) or ICA (from Citrix). The second is more universal, and in our case has more capabilities. To use it, Citrix MetaFrame software must be installed on the server. Connecting the terminal is possible either via local network Ethernet and via modem. I think the latter is exotic. After all, an analog modem cannot provide an acceptable speed of information transfer from the server and, accordingly, there can be no talk of comfortable work at the terminal.

The setup procedure is similar in both protocol types. At a minimum, you will need to enter the IP address of the terminal server and, possibly, the user login and password, if required automatic login into the system.

If you use the ICA protocol, you can activate traffic encryption and select operating modes of the terminal through the firewall.

After installing all required parameters The terminal setup can be considered complete. You can connect to the server and start working.

The setup procedure is one-time (for each server record) and takes no more than five minutes. Agree, this is much faster than installing the OS with all the necessary programs to a full-fledged workstation.

The settings specified in the initial setup wizard can always be changed later through the terminal setup panel. You can also set other specific options, for example, adjust the sensitivity of the mouse and keyboard,

as well as modify the display parameters and the IP address of the terminal.

You can also configure the printer connected to the terminal, set the local date and time, and configure a dial-up connection (using an analog modem).

The terminal firmware (internal OS) update service is presented in the Firmware tab. The firmware can be taken from a file or from a remote FTP server. Well, so that users do not have access to the terminal settings panel, you can set a password on it or even hide the panel itself.

As you can see, setting up the terminal is simple and does not take much time. After connecting to the server, the user finds himself in a familiar Windows environment and can start working. Most of the administrator's time will be spent configuring the terminal server itself (or servers, in the case of a large fleet of thin clients).

Testing

I believe that one of the critical parameters of the terminal’s operation is the minimum bandwidth at which working on it will be more or less comfortable. I conducted several tests, limiting the bandwidth between the server and the thin client to 4, 8, 16, 20, 24 Kbps when using the RDP protocol. Of course, it is impossible to draw objective conclusions, so what is written below can be considered my subjective perception.

It is inconvenient to work at 4 and 8 KB/sec (this is especially true for 4 KB/sec) windows open and move with a noticeable delay; while typing, the feeling of a rather large buffer in the keyboard is often created (letters do not appear immediately and not one at a time, but groups of 5-15 characters. Plus, the initial connection of the terminal to the server with a channel of 4 Kb/sec takes a significant amount of time. At 16 Kb, the above effects are almost completely eliminated, but some lag (delay) is still noticeable when dragging and opening windows. On a channel of 20 Kb/sec and above, working at the terminal is quite comfortable.

I note that, judging by the reviews third party sources, on slow lines big win in speed of operation is obtained when using the ICA protocol. It can work even on modem communication lines. Unfortunately, I did not have the opportunity to test the operation of the terminal on this protocol; this shortcoming will be eliminated in the next review.

Conclusions

Whether to give preference to full-fledged workstations or thin clients is up to each manager to decide for himself. In this article I just wanted to show that for some classes of tasks terminals are more profitable not only in terms of convenience, but also allow you to spend more economically cash allocated for computer equipment organizations.

Thin (terminal) clients do not require installation and maintenance of software, help to minimize the need to call technical specialists to the workplace, provide centralized installation and modernization of programs, eliminate the risk of exposure to malicious programs at the client system level, and allow efficient distribution of server resources among all workers in some places. In addition, thin clients have extremely low power consumption - about 5 W, which is 3% of the power consumption of a typical PC.

The implementation of terminal access systems with thin clients in workplaces can significantly reduce the total cost of ownership and increase the security and integrity of information

A complete terminal access system is not tied to any specific brand of equipment and consists of thin clients (terminals) - compact devices installed at workplaces, a terminal server for running user applications, and, finally, software that is already built into the server operating system system such as Microsoft Windows Server or Linux.

Terminals - thin clients

A thin client is a simplified computer completely implemented on a single board (or even in the housing of a network socket), carrying all the necessary components. The main task of a thin client is to provide a user interface with applications running on a terminal server. The thin client is designed to work in conjunction with a terminal server on which the necessary user applications are launched. Additional functionality of most thin clients is a built-in web browser. This makes it possible to work with network web applications regardless of the terminal server.

Terminal servers

Software

Terminal servers use system-wide software Windows, Linux, Solaris. Thin clients operate under the operating systems Windows CE, Windows XP Embedded, Linux. The thin client software is located directly in the thin client in the built-in flash memory. In entry-level models that do not have built-in memory, the software is downloaded from the server when the thin client is turned on (this technology is implemented in Aquarius thin clients). The latter option, however, is impractical if the communication channel between the server and the thin client has low bandwidth, or is charged based on the volume of transmitted traffic.

Articles about terminal systems and thin clients

Technorium supplies equipment and software for terminal systems, carries out installation work and adapts systems to the needs of specific business processes and operating conditions.