Peer-to-Peer versus a Client-Server


Every network, regardless of whether it is "peer-to-peer" or "client - server" based requires some form of special software in order to control the flow of information between the users being networked. A Network Operating System, or "NOS", is installed on each computer requiring network access. The NOS monitors, and at times controls, the exchange and flow of files, email, and other network information.

Network Operating Systems are classified according to whether they are peer-to-peer or client-server Network Operating Systems. A Peer-to-peer capable network operating system, such as Windows 95, Windows 98 and Windows for Workgroups are usually the best choices for home and small office networks. They do an excellent job of sharing applications, data, printers, and other local resources across a handful of computers. Client-Server network operating systems, such as Windows NT and Novel NetWare are better for larger scale organizations that require fast network access for video, publishing, multimedia, spreadsheet, database, and accounting operations. However, with the recent decreases in hardware costs, don't shy away from a client-server installation in your home or home-office if you feel that faster network access for such things as streaming video, video and web page publishing and database operations would make life easier for you.

Peer-to-Peer Networks:

Peer-to-peer networks allow you to connect two or more computers in order to pool their resources. Individual resources such as disk drives, CD-ROM drives, scanners and even printers are transformed into shared resources that are accessible from each of the computers.

Unlike client-server networks, where network information is stored on a centralized file server computer and then made available to large groups of workstation computers, the information stored over a peer-to-peer network is stored locally on each individual computer. Since peer-to-peer computers have their own hard disk drives that are accessible and sometimes shared by all of the computers on the peer-to-peer network, each computer acts as both a client (or node) and a server (information storage). In the diagram below, three peer-to-peer workstations are shown. Although not capable of handling the same rate of information flow that a client-server network would, all three computers can communicate directly with each other and share each other's resources.

A peer-to-peer network can be built with either 10BaseT cabling and a hub (as above) or with a thin coax backbone (10Base2). 10BaseT is best for small workgroups of 16 or fewer computers that are not separated by long distances, or for workgroups that have one or more portable computers that may be disconnected from the network from time to time.

Once the networking hardware has been installed, a peer-to-peer network software package must be installed on each of the computers. This software package allows information to be transferred back and forth between the computers, hard disks, and other devices connected to the computers or to the network when users request it. Windows 95 and Windows 98 both have networking software built into the operating system, and you can add other forms of peer-to-peer network operating software such as Artisoft LANtastic, and NetWare Lite. Frankly though, if you already have Windows 95 or Windows 98 (including Windows 98 Second Edition), there's really no need for additional networking software unless you have a software package that requires it.

Most network operating system software (such as Windows 95 and Windows 98) allows each peer-to-peer computer to determine which resources will be available for use by all other users of the remaining computers on the network. Specific hard and floppy disk drives, directories, files, printers, and all other resources can be attached or detached from the network via software. When one computers disk has been configured so that it is being shared, it will usually appear as a new or additional drive to the other computer users. As an example, if user A has an A and C drive on his computer, and user B configures his entire C drive so that it is shared, user A can map to the user B's C drive and have an A, C, and D drive (user A's D drive is actually user B's C drive). Directories operate in a similar fashion. If user A has an A & C drive, and user B configures his "C:\WINDOWS" and "C:\DOS" directories as sharable, user A can map to those directories and then have an A, C, D, and E drive (user A's D is user B's C:\WINDOWS, and E is user B's C:\DOS). Did you get all of that?

Because drives can be easily shared between peer-to-peer computers, data only needs to be stored on one computer, not two or three. As an example, let's say that three computers have Microsoft Word installed. Instead of saving documents and other data on all three machines, you can save all of the documents on one computer.

The advantages of peer-to-peer over client-server NOSs include:

 

Client-Server Networks:

In a client-server environment like Windows NT or Novell NetWare, files are stored on a centralized, high speed file server PC that is made available to client PCs. Network access speeds are usually faster than those found on peer-to-peer networks, which is reasonable given the vast numbers of clients that this architecture can support. Nearly all network services like printing and electronic mail are routed through the file server, which allows networking tasks to be tracked. Inefficient network segments can be reworked to make them faster, and users' activities can be closely monitored. Public data and applications are stored on the file server, where they are run from client PCs' locations, which makes upgrading software a simple task--network administrators can simply upgrade the applications stored on the file server, rather than having to physically upgrade each client PC.

In the client-server diagram below, the client computers are separate and subordinate to the file server. The primary applications and files used by each of the clients are stored in a common location on the file server. File servers are often set up so that each user on the network has access to his or her "own" directory, along with a range of "public" or shared directories where applications and data are stored. If the clients below want to communicate with each other, they must do so through the file server. A message from one client to another client is first sent to the file server, where it is then routed to its destination by the server. It becomes obvious then, that if you were to have tens or hundreds of client computers, a file server would be the only way to manage the often complex and most times simultaneous operations and transactions that large networks with many clients would generate.

Network Printing:

In client-server networks, network printing is normally handled in one of two ways, either by attaching a printer directly to the server (depending upon where the physical server is located) or by attaching a print server to the network. In most cases a print server consists of a small box with at least two connectors, one for a printer and another that attaches directly to the network cabling or into a hub. Depending upon the size of the network and/or the network print requirements, some print servers may have more than two ports. As example, a print server may have 2, 3, or 4 ports to support the same number of printers simultaneously. When a user sends a print job, it travels over the network cabling to the file server where it is stored. When the print server senses that the job is waiting, it moves it from the file server to its attached printer. When the job is finished, the print server returns a result message to the file server, indicating that the process is complete.

In the diagram below, a client computer sends a print job to the file server. The file server, in turn, forwards the job to the  print server, which sends it to the laser printer when it is on-line and available. Any client on the network can access the printer in this fashion, and it is fast. The print server can be placed anywhere on the network, and a network can have more than one print server, such as one in an office's accounting department, another in marketing, and so on.

It doesn't stop there though, as there are Print Servers available for peer-to-peer networks as well. They are incredibly convenient because they let you put a printer anywhere along your network even if there isn't a computer nearby. However, users often choose not to use a print-server with their peer-to-peer network, as every computer's resources are available to everyone else on the network. As an example, Sue can print a job on Jim's printer just as if  Sue had a printer attached to her computer.

In this next example below, a two computer peer-to-peer network, the printer is attached to the computer on the right "B". When the computer on the left "A" sends a print job, "A" thinks that it is printing to a printer of its own. While in actuality, the job travels from "A" over the network cables to computer "B" on the right, which stores (spools) and then prints the job in the background. The user at the computer "B" with the printer is never interrupted while his computer processes and prints the job transparently.

Remote Access & Modem Sharing:

When a client-server network needs a gateway to the world, the network administrator in some cases installs a remote-node server, which serves a dual function, providing remote access as well as modem sharing. Most remote-node servers attach directly to the network cabling and provide a bridge between the network, a modem, and a telephone line. Obviously, this is only one example, and there are a number of methods that can be utilized to provide the network with Internet access, including a server assigned for that specific purpose.

Remote access allows users to dial into a network from anywhere in the world. Once a connection has been established over ordinary phone lines by modem, users can access programs or data on the network just as if they were seated at one the local workstations on the network. Some remote access servers only provide access to a file server's disk drives, while others provide access to both the file server as well as direct access to any computers hard disk on the network (presuming that it has been shared). This saves time because it allows a remote user to communicate directly with any network user without having to go through the file server.

Modem sharing allows local network users to dial out from their individual network computers to access the Internet, on-line bulletin boards and much, much more. After starting up their favorite communications software, local users then establish a link with the remote-node server over the network, which then opens up an outgoing telephone line. Users' individual PCs do not need modems, which can be a money saving alternative. Only a single modem and phone line are all that is required for tens or hundreds of users. In the case of peer-to-peer networks however, every PC requires its own modem for access to the outside world unless you're using something like Microsoft's Windows® 98 Second Edition, which permits modem sharing.

Let's look at cabling next!

 

Copyright ©2000 DEW Associates Corporation. All rights reserved.
Your privacy will be respected at all times.
Year 2000 Readiness Disclosure.