Computer networks play a dominant role in transmitting information within CPA firms and other companies. A network is simply a set of computers (or terminals) interconnected by transmission paths. These paths usually take the form of telephone lines; however, other media, such as wireless and infrared transmission, radio waves, and satellite are possible. The network serves one purpose: exchange of data between computers and terminals.
Advantages of Networks
Computer networks provide several advantages. Organizations may be geographically dispersed, with offices located all over the world. Computers at each site need to transfer and exchange data, frequently on a daily basis and sometimes even in real-time. A network provides the means to exchange such data.
Even if the organization is not geographically dispersed and only has one office, networks can serve useful functions. Networks permit efficient sharing of resources. For example, if there is too much work at one site, the network allows the work to be transferred to another computer in the network. Such load sharing enhances productivity by allowing a more even and better utilization of an organization's resources.
Backup capability is an especially important feature of networks. For instance, if one computer fails, another computer in the network can take over the load. This might be critical in certain industries such as financial institutions.
Networks can be used to provide a very flexible work environment. An organization can allow its employees to connect to the network and work from home or "telecommute." A network makes it easier for employees to travel to remote locations and still have access to critical data such as sales for last week or research data from a project.
Data flows between computers in a network using one of three methods. Simplex transmission is in one direction only. An example of simplex transmission is radio or television transmission. Simplex transmission is rare in computer networks due to the one-way nature of data transmission. Half-duplex transmission-information can flow in both directions-is found in many systems. However, it is not possible for the information to flow in both directions simultaneously. In other words, once a query is transmitted from one device, it must wait for a response to come back. A full-duplex system can transmit information in both directions simultaneously; it does not have the intervening stop-and-wait aspect of half-duplex systems. For high throughput and fast response time, full-duplex transmission is essential.
Data switching equipment is used to route data through the network to its final destinations. For instance, data switching equipment is used to route data around failed or busy devices or channels.
In designing the network, three factors must be considered. First, the user should get the best response time and throughput. This is especially important for interactive sessions between user applications. Throughput involves transmitting the maximum amount of data per unit of time.
Second, the data should be transmitted along the least-cost path within the network, as long as other factors, such as reliability, are not compromised. The leastcost path is generally the shortest channel between devices and involves the use of the fewest number of intermediate components. Furthermore, low priority data can be transmitted over relatively inexpensive telephone lines, while high priority data can be transmitted over expensive high speed satellite channels.
Third, maximum reliability should be provided to assure proper receipt of all data traffic. Network reliability includes not only the ability to deliver error-free data but also the ability to recover from errors or lost data in the network. The network's diagnostic system should be capable of locating problems with components and perhaps even isolating the component from the network.
A good communications program will have numerous protocol options, enabling communications with different types of equipment. Some communications programs do error checking of information on software programs received. Desirable features in telecommunications programs include menus providing help, telephone directory storage, and automatic log-on and redial.
The network configuration or topology is the physical shape of the network in terms of the layout of linking stations. A node refers to a workstation. A bridge is a connection between two similar networks. Network protocols are software implementations providing support for network data transmission. A server is a micro or a peripheral performing tasks such as data storage functions within a local area network (LAN). Network servers are of several types. A dedicated server is a central computer used only to manage network traffic. A computer that is used simultaneously as a local workstation is called a nondedicated server. In general, dedicated servers provide faster network performance since they do not take requests from both local users and network stations. In addition, these machines are not susceptible to crashes caused by local users' errors. Dedicated servers are expensive and cannot be disconnected from the network and used as stand-alone computers. Nondedicated servers have a higher price-performance ratio for companies that occasionally use the server as a local workstation.
The most common types of network topologies are shown in the Exhibits and are as follows:
The hierarchical topology (also called vertical or tree structure) is one of the most common networks. The hierarchical topology is attractive for several reasons. The software to control the network is simple, and the topology provides a concentration point for control and error resolution. However, it also presents a potential bottleneck and reliability problems. It is possible that network capabilities may be completely lost in the event of a failure at a higher level.
The horizontal topology (or bus topology) is popular in local area networks. Its advantages include simple traffic flow between devices. This topology permits all devices to receive every transmission; a single station broadcasts to multiple stations. The biggest disadvantage is that since all computers share a single channel, a failure in the communication channel results in the loss of the network. One way to get around this problem is through the use of redundant channels. Another disadvantage with this topology is that the absence of concentration points makes it more difficult to isolate faults to any particular component. Identifying a problem requires the checking of each system element. A bus topology is suggested for shared databases but is not good for single-message switching.
The star topology is very popular and is widely used for data communication systems. The software for star topology is not complex, and controlling traffic is simple. All traffic emanates from the hub or the center of the star. In a way, the star configuration is similar to the hierarchical network; however, the star topology has more limited distributed processing capabilities. The hub is responsible for routing data traffic to other components. It is also responsible for isolating faults, which is a relatively simple matter in the star configuration. The star network, like the hierarchical network, is subject to a potential bottleneck at the hub and may cause serious reliability problems. One way to minimize this problem and enhance reliability is by establishing a redundant backup of the hub node. A star network is best when there is a need to enter and process data at many locations with day-end distribution to different remote users. Here, information for general use is sent to the host computer for subsequent processing. It is easy to identify errors in the system, since each communication must go through the central controller. While maintenance is easily conducted, if the central computer fails, the network stops. There is a high initial cost in setting up the system because each node requires hookup to the host computer in addition to the host's cost. Expansion is easy; all that is needed is to run a wire from the terminal to the host computer.
The ring topology is another popular approach to structuring a network. The data in a ring network flows in the circular direction, usually in one direction only. The data flows from one station to the next; each station receives the data and then transmits it to the next station. One main advantage of the ring network is that bottlenecks, such as those found in the hierarchical or star networks, are relatively uncommon. There is an organized structure. The primary disadvantage of the ring network is that a single channel ties all of the components in a network. The entire network can be lost if the channel between two nodes fails. This problem can usually be alleviated by establishing a backup channel. Other ways to overcome this problem are using switches to automatically route the traffic around the failed node or installing redundant cables. A ring network is more reliable and less expensive when there is a minimum level of communication between micros. This type of the network is best when there are several users at different locations who have to access updated data on a continual basis. Here, more than one data transmission can occur simultaneously. The ring network permits accountants within the firm to create and update shared databases.
The mesh topology provides a very reliable, though complex, network. Its structure makes it relatively immune to bottlenecks and other failures. The multiplicity of paths makes it relatively easy to route traffic around failed components or busy nodes.
Wide Area Neworks and Local Arks may be Networks
Networks may be broadly classified as either wide area networks (WANs) or local area networks (LANs). The computers in a WAN may be anywhere from several miles to thousands of miles apart. In contrast, the computers in a LAN are usually closer together, such as in a building or a plant. Data switching equipment might be used in LANs but not as frequently as it is in WANs.
The channels in WANs are usually provided by an interchange carrier, such as AT&T or MCI, for a monthly fee plus usage cost. These channels are usually slow and relatively error-prone. In contrast, the channels in a LAN are usually fast and relatively error free.
The LAN topology tends to be more structured. Since the channels in a LAN network are relatively inexpensive, the owners of a LAN are generally not concerned with the maximum utilization of channels. Furthermore, since LANS usually reside in a building or a plant, such networks tend to be inherently more structured and ordered. LANs are flexible, fast, compatible, maximize equipment utilization, and reduce processing cost and errors while providing ease of information flow. LANs use ordinary telephone lines, coaxial cables, and fiber optics. Fiber optics result in good performance and reliability, but have a high cost. LAN performance depends on physical design, protocols supported, and transmission bandwidth. Bandwidth is the frequency range of a channel and reflects transmission speed along the network. Transmission speed is slowed down as more devices become part of the LAN.
Two or more LANs may be interconnected. Each node becomes a cluster of stations (subnetworks). The LANs communicate with each other.
A network user should be positively authenticated before allowing access to the system. Different levels of access privileges should exist depending upon the security clearance level. Access controls should also apply to terminals and applications. A record should be kept of date, time, and files accessed. Improper use should "lock" the terminal.
Cryptography (coded messages) must be used to provide a secure communication link between interconnected computers. The objective is to make transmitted data unintelligible to unauthorized users while permitting decoding to those having authorized access.
Security controls must exist over the host computers and subnetworks. Security should be in different layers depending on confidentiality and importance. Security aspects of each network must be distributed to the gateways so as to incorporate security and controls in routing decisions. Communication security may be in the form of the following:
Access control guards against improper use of the network. For example, KERBEROS is authentication software that is added to an existing security system to verify a user's existence and assure that the user is not an impostor. KERBEROS does this by encrypting passwords transmitted through networks.
Password control and user authentication devices may be used such as Security Dynamics Secur ID (800-SECURID) and VASCO Data Security's Access Key II [(800) 238-2726]. Networks should not accept a prepaid call if it is not from a network user. Hackers typically do not spend their own funds. Other security steps include reviewing data communication billings, verifying each host-to-host connection, reviewing all dial-up terminal users, and making all the telephone numbers unlisted with periodic changes thereto.
Identfication identifies the origin of a communication within the network such as identifying the entity involved using digital signals or notarization.
Data confidentiality maintains confidentiality over unauthorized disclosure of information within the communication process.
Data integrity guards against unauthorized changes (e.g., adding, deleting) of data at both the receiving and sending points such as through cryptographic methods. Anti-virus software should be installed at both the network server and workstations.
Authentication substantiates the identity of an originating or user entity within the network. There is verification that the entity is actually the one being claimed and that the information being transmitted is appropriate. Examples of security controls are passwords, time stamping, synchronized checks, nonrepudiation, and multiple-way handshakes.
Digital signature messages are signed with a private key.
Routing control inhibits data flow to insecure network elements such as identified insecure subnetworks.
Traffic padding performs a traffic analysis of data for reasonableness.
There are various ways to back up data in networks. For a small network, one workstation may be used as the backup and restore for other nodes. In a large network, backup may be done by several servers since the failure of one could have disastrous effects on the entire system. Additional backup copies should be kept in a secure off-site location. Access to backup files must be strictly controlled.
The Future of Networking
At the present time, corporate networking is primarily used for accessing or sending word processing documents, spreadsheets, and database applications in a fairly routine utilitarian fashion. Clearly, networking is about to undergo a dramatic transformation as the advance of software and connectivity to the Internet is ring multimedia available to the corporate user. Not only will multimedia presentations be fully accessible, but the user can become fully interactive with them.
Thus, future networking will be based on highly sophisticated multimedia enabled computers. While the end product of today's corporate network is a printed document, future networked computing will allow hands-free video conferencing using small video cameras and microphones connected directly to a computer workstation. These user friendly computer workstations will be seamlessly integrated into the work environment.
Multimedia network computer applications will allow the accountant to develop sophisticated, creative, and interactive financial presentations utilizing color, sound, and motion. These presentations will not only be the current printed documents, but informational videos also will be available on demand from menu-driven videoservers.
Bandwidth is the major technological challenge for the future development of computer networks. It determines the speed data can be moved through a network. The common Ethernet technology being employed on the overwhelming majority of today's networks moves data at approximately 10 mbps, while the widely used token ring topology moves it at 16 mbps. These speeds have not changed in over 10 years. Relative to the development of today's high speed CPUs, these slow network transmission speeds are the greatest challenge to the future of organizational networking. The need for faster networks is imperative; however, there are several competing technologies.
One promising solution for corporate networks is fast Ethernet. It runs at 100 mbps, which is 10 times faster than regular Ethernet. The cost of the required fast Ethernet interface cards has fallen to just a small premium over standard Ethernet interface cards. A major component of fast Ethernet is the Ethernet switch which replaces current network bridges and routers by creating discrete connections between network nodes.
For much more data intensive LANS, such as carriers of large graphics, database files, or video streams, there are two choices. These are asynchronous transfer mode (ATM) or fibre channel. ATM is a highly adaptable "fast packet" emerging technology that is gaining pop ularity at the WAN level, but is less successful at the LAN level because of a current lack of standardization. Fibre channel is a developing technology based on fiber-optic cabling. This technology was first defined by the American National Standards Institute (ANSI). Subsequently, it has been refined through the development of better and cheaper cable manufacturing methods and new, sophisticated network switching techniques. Fibre channel is a fast and reliable network technology which appears to have a bright future.
The development of WANS by connecting LANS over telephone company lines is currently the fastest development in networking. A rapidly spreading development in WAN technology is sending voice signals through a data communication network, effectively bypassing expensive long distance telephone charges: The technology works. One company, Micom, recently purchased by NorTel, received the "product of the year" award from Lan Magazine for its voice capable frame relay network router.
Integrated services digital network (ISDN) is another rapidly developing technology in WANS. ISDN combines voice and digital network services in a single medium, making it possible to offer digital data services as well as voice connections through a single "wire." However, while ISDN offers much wider bandwidth and speed than conventional analog technology, its relative speed advantage pales in comparison to ATM. ATM guidelines have already been defined for the simultaneous transmission of data (audio and video) at speeds in the gigabite per second range. One company, FORE systems, has developed ATM routers and switches-which can connect two LANS together, either directly or by using telephone company lines, and achieve blindingly fast WAN connectivity.
There is no question that advances in computer network technology will play a major role in future accounting and business relationships. Multimedia, full motion video and audio will soon be the norm in network applications. Increasingly, we will see hands-down computing occur where direct face-to-face desktop communication around the world will be as seamless as using a basic intercom system.…