LINKING IN COMMUNICATIONS NETWORKS

Linking involves two types of paths through the network: physical and virtual. Paths are created through the network when physical links are connected in series between two terminal sites. A receive path output from equipment in a terminal is connected to transmit path input on another facility. For a full-duplex or two-way path, the receive path from the opposite direction must be connected to the transmit path in the opposite direction. This type of linking process in digital networks can be extended many times without undue service impairment, except for the accumulation of transmission delay and errors. The latter can be mitigated by good link engineering practice common in radio and optical link budgeting.

When any network is made up of three or more sites, another term comes in to play called meshing. Networks are either fully meshed or partially meshed. The four-site network depicted in Figure 1 is classified as partially meshed because there is no direct path or link between sites 1 and 3 or between sites 2 and 4. Figure 2 shows a fully meshed four-site network.

Figure 1: Four-Site Fully Meshed Network

The difference in meshing has implications in terms of economics, reliability, and robustness. Economically, the partially meshed network connected in a round-robin fashion requires four links. A fully meshed network would require six links. The two additional links might increase the monthly charges for leased private line facilities by 50% when comparing a fully meshed, four-site network to a partially meshed network. Such a choice or decision is the eternal dilemma of network architects and designers. The way out of the woods requires economic analysis and judgment to resolve. For now, simply note that two-site networks need not be concerned about meshing issues. Three-site networks require four links to be fully meshed, four-site networks require six links to fully mesh, and a five-site network requires 10 links. Figure 2 shows a five-site fully meshed network.

Figure 2: Five-Site Fully Meshed Network

As the number of sites in a network increase, the number of links soars, dramatically increasing complexity. The more complex network architecture becomes, the greater the need for more detailed documentation required to manage cost, reliability, robustness, and network performance. Another factor mitigating the need for fully meshed networks is use. For example, consider that when one site connects to another, it is for the purpose of passing traffic. That traffic may be flow only one way or both ways, and if the equipment or people at the site are busy or engaged with one site, it may not be possible to connect to a third site simultaneously. From a slightly different perspective, if two sites are busy, then the likelihood of either of them becoming engaged with a third site is limited. Therefore, use of the other possible links is significantly less likely, so why spend money on all that capacity. However, without a fully meshed network, ways to communicate between all the sites must be established, and the answer is switching. The carrier might provide the switching function, or it can be done with premises equipment.

One last point is that meshing might occur at any one of the communications layers, all the way from the physical layer through and including layer 4 of the OSI stack.

Virtual paths at layer 1 are created in the electrical domain, across disparate physical facilities. Linking of virtual paths involves a combination of several factors such as physical aspects of the connector, a match between transmit and receive pins or pairs, signal polarity, and clocking. For example, digital cross connect switching systems are used to create a virtual path through the transmission network when provisioning private line services. POTS/ISDN switches and their signaling systems create virtual paths through the same transmission network to support voice grade connections enabling telephone calls, facsimile, and modem transmission. The mechanics of the connection involve connecting a transmit/receive pair on one side of the switching system to a receive/transmit pair on another facility.

ATM network architecture includes a virtual path layer, inside which virtual channels or circuits are created and placed. Routing and switching are performed according to information in the virtual path identifier and virtual circuit identifier sections of the ATM cell header.

The linking term is also applied to a process or protocol to create a path for data or information between disparate media. A link created with IEEE 802.2 logical link control (LLC) at layer 2 is at the highest layer of LAN architecture. It defines a set of protocols that support services between the media access layer and the transport layer. LLC is functionally equivalent to the telephone hook-switch and DTMF keypad used to control setup and teardown of a voice connection or link. Most LAN cards, and many other devices supporting LAN connections, have a green indicator light. If the light is illuminated, it indicates physical connectivity between two devices. Many telephones, especially those with two or more lines, have an indicator to show off-hook, active line, or in use. Most equipment with a wide area physical connection is equipped with some kind of indicator as well as alarm to indicate status of the link.

LAYERING AS USED IN COMPUTERS AND COMMUNICATIONS NETWORKS

The basic idea behind layering is that computer equipment and system functions are bound by, or within, layers. Each layer is bound to, or interacts with, its neighbor immediately above and below. If each layer up and down the stack interacts, or interoperates, with its neighbor successfully, then the system or process making up the overall system or network is likely to succeed in performing all the functions it was designed to accomplish.

The layering concept was created by the International Standards Organization to serve as a standard definition of computer industry structure dealing with communications issues in computer environments. The standard is named the open systems interconnect (OSI) model and typically referred to as the OSI stack. The communications industry adopted the technique for use in standards and design documents. Two examples of communications stacks are shown in Figure 1 along with the OSI stack. The OSI stack has seven layers; the synchronous optical network/synchronous digital hierarchy (SONET/SDH) and Internet stacks are both four layers.

Figure 1: Layering Models Used in Computing and Communications

Taken in isolated display, there doesn’t appear to be much of a match or direct relationship between the three. However, we know that SONET/SDH is an example of the physical layer of the OSI stack, and only the physical layer.

The four-layer Internet stack used by many rests on an IP layer, which is a peer to the network layer in the OSI stack.

Each of the layers in the OSI stack is unique in function and behavior, and has a specific relationship with its neighbors above and below. The OSI model is subdivided with layer 1 and 2 classified as being hardware-oriented, whereas the upper layers are said to be software-oriented. As an entity, the lower layers are communications-oriented; the upper layers are user- or application-oriented.

Another commonly used way of illustrating the layering concept and relating it directly to linking is to draw an end-to-end service model with multiple layers showing multiple virtual channels on top of a physical channel. Figure 2 is another example of the use of layering models to show relationships between the layers in communications networks and computers.

Figure 2: Example of Three Layering Models

Figure 2 has been redrawn from the three separate models shown in Figure 1. It shows these three models and how they relate to each other.

The OSI model has been separated between layer 2 and layer 3. This is the division between hardware on the lower level and software on the higher level. The Internet model is left intact and placed above the SONET model. The VT layer has been added to the SONET model.

Layers 2, 3, and 4 have been shaded with four different levels or shades of gray, reflecting the dividing layer between hardware and software in the OSI model. There is a similar division between the IP, and lowest layer in the Internet model and the VT/STS, or highest layer, in the SONET/SDH model. The data link layer is a point of commonality across all three. That is, voice and data networks use exactly the same physical layer network interfaces and protocols as the Internet does. For example, an unchannelized E1/T1, E3/DS3, or OC3/STM1 facility can be used to support all types of traffic defined in the site location architecture.

Most depictions of SONET/SDH layering don’t include the virtual tributary (VT) layer. If this layer is included on top of the SONET/SDH model, it becomes a direct fit, or interface to the Internet model for plesiochronous (PDH) point-to-point links or leased/private line facilities. Another point worth mentioning is the fact that many purveyors of network equipment and services are offering direct interface between IP and SONET/SDH synchronous transport streams. In and of itself, it’s an incremental step. However, if this physical layer capability is combined with differentiated services in the network, traffic aggregation, and type of service (TOS) capabilities in new and emerging network equipment, the result is a potentially dramatic and profound impact on ISDN/PSTN and plesiochronous (PDH) network facilities and services.

Consolidating mixed or disparate traffic requires knowledge of all three layers and a detailed understanding of how they relate to each other when integrating equipment and software. More importantly, understanding how the traffic payloads are organized and structured is key to successfully mapping the traffic to the network to get maximum use of the network. Much has been said and written about convergence or converged networks. Occasionally, the term multi-service network is used. Layer 2 and 3 is the place in the layering models where multiple, disparate traffic types are converged and mapped to a common access and/or transport facility. Physical placement of routers and switching equipment and its configuration determines physically where, and in what sequence, the so-called convergence (of disparate traffic) takes place.

Applying the layering concept to communications networks seems quite natural and logical. Two types of layering are commonly used to depict communications networks. Classical telephone networks are structured around a multiplexing, switching, and transport hierarchy, while computer networking and the Internet are structured around protocols and interfaces.

Another term that creeps into the lingo from time to time is overlay network. For example, the larger multi-service carrier networks share transmission facilities between voice grade services that require 64 Kbs transmission links with ATM switches and IP routers that use raw transmission bandwidth in varying amounts. Since the ATM and IP networks came into existence after voice grade services, they were built and are said to overlay voice grade services. Occasionally this term is also understood to mean the ATM and IP networks are above voice grade services in the OSI stack, where voice grade services are seen at layer 2, ATM at layer 2/3, and IP at layer 3/4.

Continuing the evolution, the Internet community has taken these two separate models and created a separate but related structure dubbed Inter-networks. The Internet is a completely different structure with its own unique behavior in terms of how it moves a payload, otherwise called content or information. The concept is built around an idea that combines payload information with address information, hands it to the network, and the network not only carries the information, but when it gets to its destination, the network communicates that fact back to the sender. Overall result: A third network is built using the same kinds of standard layer 1 and layer 2 facilities as voice and data networks.

Continuing to add communications facilities for separate network applications has led many to question and wonder if there may not be a better way to organize their traffic to get better use of all the facilities being paid for. Attempts to address these issues have led to solutions called converged or multi-service networks. A more appropriate label might start with the traffic whereby disparate traffic types are converged into a common access facility or across a common transport facility. Viewing the three layered models in a single context can be a constructive and instructive step to defining requirements for a network capable of carrying disparate traffic.

Enterprise Networks | Environments

Enterprise networks are used by businesses to accomplish the everyday functions essential to their success. They are characterized by requirements for features, functionality, and scale that go far beyond typical consumer needs for telecommunications services. Enterprise networks can be very large, spanning the entire globe and operating with multimillion-dollar annual budgets, or they can be much more modest in scale, conforming to the needs of a small business operating in a distinct geographical area.

No matter how large or small, one defining characteristic of an enterprise network is that in almost all cases it is a means to another end. That is, the company that pays for and supports the network is not primarily in the communications business, but rather is in some other business supported by the network. Salespeople make calls to customers and to the home office. Product planners on opposite sides of a continent or an ocean hold weekly conference calls supported by high-quality graphics. A call center takes customer orders or provides support to users of the firm’s products. Remote workers dial in to the company network for e-mail and voice mail. All of these activities are essential to conducting the firm’s business, and some may place critical and demanding requirements on the network, but they are not in themselves the reason why the firm is in business or the sources of its revenue.

The degree to which the network is managed and operated by the firm itself, or to which these functions are outsourced to another company, varies from firm to firm. In the one case, you as the network manager may have a staff of employees that you hire, compensate, and coach in performing tasks to keep the network up and running and expand and extend it to keep pace with user demand. In the other case, you may be primarily responsible for managing the relationship with an outsourcing vendor who in turn performs these tasks. Either way, one of your central concerns is likely to be the costs of network operations, and managing these costs within a budget may be a major function of your job.

Enterprise networks often differ in scale and geographical scope, as mentioned previously. You must consider size and scale when you evaluate your Internet and telecommunications integration needs. Since cost savings is an important goal, it is essential to understand how scale and geographical scope can affect IP telephony costs, including the effects of government regulation in different jurisdictions.

The following sections introduce some possible enterprise network environments. As noted previously, you may find that your own environment is some combination of these, or that it is somewhat different from any of them. Nonetheless, they should provide a framework for thinking about the applicability of Internet and telecommunications integration to your business.

Small Office/Home Office (SOHO)

A SOHO environment can be a small office in a business location, a small office located in a private home, or even a home office setup designed primarily to support a remote worker for a large business. There may be several workers in the office with voice and data communication needs, or there may be just one. In either case, needs for communication within the office are likely to be minimal; the most demanding need is for communication outside the office with customers and suppliers, and, in the remote worker case, with other employees and centralized facilities of the large business. Depending on the nature of the business, the scope of communication may be local, regional, national, or international.

To support voice communications, the most readily available traditional options will be based on switched telephone service from the local telephony provider, possibly augmented by relatively simple premises telephone equipment (an electronic key system or the like) to help manage multiple lines. Connections to Internet service providers (ISPs) or the internal network of a large business can also most readily be supported on a dial-up basis. Depending on the locality, you may have other options such as cable modems or digital subscriber line (DSL) technology.

Medium to Large Business—Single Building or Campus

The medium to large business with a single building or campus category can span locations with less than a dozen to more than 1000 employees. A defining characteristic is a significant need to provide communication among the colocated employees (for example, at an R&D site or large professional office) and/or to manage high-volume incoming or outgoing communications (as at a call center).

Traditionally, voice communications for a single campus location would be handled by a digital private branch exchange (PBX) supported by a substantial in-building twisted pair wiring infrastructure. In a minority of cases (usually governmental organizations or other institutions with a limited ability to raise capital for buying things like PBXs), the same functions may be provided by Centrex service from a telephone company. For call center operations, a special feature of PBX called automatic call distribution (ACD) may be in use, possibly along with computer telephony (CT) capabilities to optimize the flow of calls and connect agents with customer information databases and other information systems. Data communications would be provided by local area network (LAN) technology, probably running over the same twisted pair infrastructure as the telephone system, using networking technology like switches and routers. Depending on the scale of the operation, you may have a hierarchy of departmental and backbone LANs.

Communications outside the building or campus may use a combination of local, regional, national, and international services, depending on the nature of the business. If the building or campus is part of a larger business, a substantial amount of off-premises voice and data traffic may flow over private facilities such as leased circuits or specialized carrier services like software defined networks (primarily for voice) or frame relay service (traditionally for data). Switched business lines from the local telephony provider are available as the ultimate default.

Medium to Large Business—Multiple Campuses

The medium to large business with more than one campus is a generalization of the preceding category. All of the same characteristics are present. The new factor is that there are substantial traffic flows between two or more buildings/campuses. As a result, there will often be significant use of private facilities or specialized carrier services to connect the multiple large locations. Distances between pairs of buildings/campuses, of course, can vary enormously, from a few miles/kilometers to intercontinental distances, and may have a great impact on the specific services available for interconnection as well as on the relative economics of different options.

Medium to Large Business—Single or Multiple Campuses with Small Remote Locations

Many medium to large organizations fall in the category of medium to large businesses with single or multiple campuses with small remote locations. Not only are there possibly multiple large buildings or campuses with substantial traffic flows among them, but there are very important communication flows to and from a relatively large number of small locations. These small locations may be similar in character to the small office/home office environments described previously, or they may be a bit larger, more like the lower end of the single-campus category. All of the communication options described in the preceding two categories may be in place, and in addition the small remote offices present important communication requirements that may need to be satisfied by a combination of switched business lines, specialized carrier services, and, if traffic densities are high enough and the economics work out, leased lines like T1s or cable-based circuits.

The limiting case of the small remote office is probably the single nomadic worker who needs to stay in touch with the business through whatever combination of e-mail, voice mail, and fax he or she can manage. Communications options for this increasingly important case are typically based on dial-up from wired or wireless telephones into corporate voice networks and/or remote access servers for corporate data networks.

Interenterprise Networks and Extranets

All of the preceding categories include some amount of communication between the enterprise in question and other businesses, including suppliers, distributors, and business customers. The default communication mechanism for such communication flows is the public telephone network, whether employed for voice or fax transmission, or maybe the post office, or, in a small but growing number of instances, the public Internet. However, for high-volume transactions between companies that are regular, long-term partners, intercompany private networks or industry-wide jointly managed networks are sometimes employed. Examples are the use of electronic data interchange (EDI) for functions like ordering and invoicing, and electronic funds transfers in the banking industry. Internet technology is often adapted to interenterprise uses in the form of an extranet—an IP-based network that is neither public nor strictly private and that is used by two or more businesses with a long-term relationship.