[attachment=13735]
Introduction to Remote LAN Networks
Not very long ago, remote access to enterprise networks was like caviar, luxury limousines and upper class beach resorts- available only to a privileged few. The concept of working at home was virtually unknown. Connections to home offices, where they existed, were extremely limited in number and scope. They were mostly restricted to insurance offices, field sales offices, and a few well-to-do lawyers. For almost everyone else, work was synonymous with a daily commute to the office.
Even connections to corporate computing services from branch offices were far from pervasive. In fact, the very subject of remote connectivity was avoided as much as possible by network managers and their staffs. The topic simply raised too many thorny technical, financial and political issues. Almost without exception, every part of remote access presented a barrier of some sort to implementation. However, the two factors that worked the hardest against remote access were logistics and economics.
Logistics came into play because it was not a straightforward matter to provide connections to remote locations. Installations were almost always a logistical nightmare. In addition, installations involved a fair amount of work and coordination, both by the telephone company and by corporate information services.
Remote connectivity costs were as astronomical as installations were complex. Equipment was expensive, line costs were far from inexpensive and the personnel required to support and service remote connections did not come cheaply.
Where remote connections did occur, the most common form of access was a slow-speed serial page link to the infamous dumb terminal. These terminals were rather large, intimidating devices that were usually connected to the corporate mainframe by coaxial cable. They also required a rather odd-looking device called a controller.
Figure 1.1 shows how remote connections were typically accomplished. The mainframe was channel-attached to a front end processor (FEP) in the computer room. The FEP connected to a synchronous or limited-distance modem. A synchronous page link was used to connect the FEP to the remote controller.
Sometimes a bank of synchronous modems or limited-distance modems connected a number of remote terminals to the FEP. In either case, a dedicated page link connected the remote destination through the telephone company's central office. The page link was channeled through the central office, but did not usually go through the telephone-switching equipment. At the destination was another synchronous or limited-distance modem. This modem connected to the controller with a serial cable. Finally, coax connected the terminal with the controller.
To the user the connection seemed quite ordinary. The terminal appeared to be channel-attached directly to the FEP. Anything that could ordinarily be done on a terminal in the corporate office could also be done on the remote terminal in the branch office.
To the information services staff, the connection did not appear to be ordinary at all. The word that a terminal had to be set up in a remote office location often sent chills down the corporate information systems manager's spine.
It was a job in every sense of the word. Terminals and controllers had to be shipped to the remote site, configured and installed. A data line had to be run from the remote site to the corporate computing center. The telephone company had to dedicate wire pairs in their cables to the connection.
Once connections were in place, they were difficult to change. Moving the location of the terminal required changes that were neither cheap nor easy. Relocating remote circuits required the generation of telephone company work orders. This took weeks to accomplish and usually required several visits to the remote locations. The process was so costly and involved that relocating remote equipment was strongly discouraged. Eventually, additions and changes became nearly impossible to accommodate as open copper pairs in the telephone company's cable plants became scarce.
Figure 1.1 Branch office connections through dedicated links
The Transition to Local Area Networks
Today things have changed remarkably. Within a building or corporate site, communications capabilities have undergone a radical transformation. The prime mover for change was the deployment of Local Area Network (LAN) technology.
LANs have radically changed the corporate information infrastructure. They have led corporations to move away from terminal- and mainframe-based systems. These have been replaced with client/server-based information systems connected over LANs. Terminals were replaced by personal computers that could either emulate a terminal or become an intelligent part of the network in their own right. In many cases, mainframes became orphans and were replaced by network servers.
Where mainframes remain in service, their role has been often been relegated to handling large file or mail services. The mainframes themselves are, more often than not, accessed through LANs. All of this has given rise to new capabilities for information access.
Popular LANs
Few developments in technology have made such quick inroads into corporate computing environments as Local Area Networks. They have become extremely popular, and with very good reason.
LANs provided an overall level of connectivity that surpassed anything previously available. They offered throughput rates that were adequate for even the most bandwidth-hungry applications. LANs allowed users to transcend the limitations of their local computer, and to access local, regional and worldwide information networks.
LANs provide high-speed multiple-user access to a wide variety of information sources residing on devices such as file servers, minicomputers and mainframes. Peripheral devices such as printers, modems, and fax systems, can be connected via a LAN, making them accessible to multiple users. Today, virtually no one questions the statement that LANs were the most important technological advancement made during the last half of the eighties.
Limited LANs
However, with all their capabilities, Local Area Networks are far from perfect. They have limitations, sometimes significantly so, particularly in offering widespread connectivity. It is interesting that the very name given to LANs highlights their biggest limitation. Local Area Networks are just that- local. They are designed to provide network services only within a restricted area. This was not an oversight on the part of the LAN designers, it is exactly what the LAN designers intended.
Local Area Networks are local for a good reason. Early LAN designers never envisioned that their networks would someday be extended outside a fairly small radius. As a result, LAN technologies such as Ethernet and Token Ring deployed technology that has distance limitations. The intent was to use the LAN for network connections on the same floor or within the same building.
Distance causes erosion of the digital signal that is used to propagate information over the LAN. Besides signal loss issues, there are also delay problems with long connections. This is particularly true with Ethernet, since it uses a bus topology to connect workstations. Furthermore, the algorithm used to carry Ethernet causes the network to be even more limited when it comes to distance.
Devices on an Ethernet network use an algorithm called Carrier Sense Multiple Access with Collision Detection (CSMA/CD). The CSMA/CD algorithm listens on the wire before it transmits in case another station happens to be transmitting at the exact same moment. Propagation delays along a lengthy backbone cable cause stations that are physically far apart to transmit at the exact same moment. The length of backbone cable masks the fact that more than one station is transmitting. This causes collisions to occur on the network. Therefore, Ethernet over thinnet or twisted-pair cable carries a distance limitation of 100 meters. This limits the propagation delay and helps prevent collisions from occurring on the network.
To get beyond the distance limitations required LAN segmentation and devices called repeaters. However, even this does not allow unlimited LAN distances. Therefore, LANs rarely traversed building boundaries. It was even less common for them to leave a corporate or university campus.