Introduction
A WAN, also known as a long-haul network is a loose term used to distinguish networking technologies from a geographical perspective. In contrast to local area networks (LAN), WANs provide communications over substantially longer distances. Note that the definition of long-distance is a vague term that could entail crossing a state, a country or even an ocean. In contrast to LANs where organizations typically own and manage their network, WAN services are typically a pay-for-service and are provided and managed by regional telecommunications carriers.
As previously cited, a LAN’s primary purpose is to provide network access to hosts using various topologies. In contrast to LANs, WANs are designed to interconnect networks through serial point-to-point links. Recall a physical point-to-point topology connects two nodes directly. In a point-to-point WAN architecture, layer two frames are placed on the communications medium at the source and removed at the destination. This architecture allows the data link MAC protocols to be very simple since frames can only travel between the two nodes. This simplicity allows the layer two switching to be optimized and performed in hardware achieving an efficient low latency communications channel. Lastly, it should be noted that in contrast to the quick speeds seen in today’s LANs, WANs historically operate at slower speeds and have a greater delay between connections.
With this basis, WANs may be characterized as a series of specialized computers also known as packet switching nodes (PSN) that connect networks through serial long haul communications lines . In summary, WANs may be distinguished from LANs based on the following three major characteristics: (a) WANs connect networks across wide geographic areas, (b) WANs typically use the services of large network providers that include telephone companies, cable companies and satellite systems, and (c) WANs use serial point-to-point connections . Implicit in this serial communications architecture is the need to share the communications channel and accommodate concurrent dialogs through multiplexing.
WAN Analysis Basis
To serve as a foundation for analysis, WANs can be distinguished based on the following characteristics: (a) packet switched or circuit switched connections, (b) synchronous or asynchronous communications, (c) narrowband or broadband capacity, (d) end-to-end delivery or used solely as an intermediate transport network, (e) dedicated or on-demand connections, and (f) the types of communications media (e.g. fiber optic or co-axial cable). Packet switching and circuit switching will be discussed in detail below. Synchronous serial connections provide synchronization through an external clock. This requires that the synchronous transmissions contain a separate or second signal that allows the destination to discern where transmissions begin and end. Synchronous transmission clocking will be discussed below in the WAN devices section analyzing data communications equipment. Asynchronous serial connections embed the clocking in the signal as previously examined in earlier (e.g. Manchester Encoding).
Narrowband and broadband indicate the communication medium’s capacity. The distinction between narrowband and broadband is also vague however Cisco (2004) states that the demarcation is 128 Kilobits per second (Kbps). End-to-end delivery can be further distinguished between dedicated end-to-end point-to-point connections, permanent switched circuits (PVC) and virtual switched circuits (SVC). Intermediary transport links are the serial point-to-point links within the WAN. Dedicated circuits provide consistent always on bandwidth whereas on-demand services are provided in response to need. To provide a sound foundation for analysis, it is worthwhile to provide a historical context and trace the history of the largest and most well known WAN – the Internet.
WAN History
The Internet can trace its history to the ARPANET. ARPANET was founded by the Advanced Research Projects Agency (ARPA) in 1968 to serve as test bed for packet switching technology . Implicit in this packet switching technology is the multiplexed communications architecture introduced above. An important design goal of ARPANET was to create a resilient network that was not susceptible a single point of failure. To achieve this goal, ARPANET specified a redundant partial mesh topology where packets could dynamically take different routes as necessary. Shortly after it its inception, ARPANET evolved to consist of roughly 210 PSNs that were minicomputers interconnected point-to-point . ARPANET quickly demonstrated that it had become a dependable backbone and in 1975, the Department of Defense assumed control of ARPANET. Note there were several other government agencies involved in this early Internetwork development that notably includes the National Science Foundation (NSF) and their development of NSFNET versions one and two.
It could be asserted the transition from government development and management of the Internet to private development and management occurred in 1995 when NSF awarded MCI a contract to build a 155 Mb per second (Mbps) speed backbone. Since this time Internet development has increasingly been a proprietary endeavor receiving fewer and fewer funds from the federal government . With this basis as introduced above, the Internet and WAN connectivity have largely become charge-for-service networks .
During this evolution from ARPANET to today’s Internet, the industry observed the emergence and evolution of many technologies. These emergent technologies include: (a) point-to-point leased lines, (b) packet-switched networks, and (c) circuit-switched networks . Examples of these technologies listed respectively include: (a) digital subscriber lines (DSL), cable and leased line point-to-point networks, (b) X.25, Frame Relay, Asynchronous Transfer Mode (ATM) and the emergent Multiprotocol Label Switching (MPLS) packet switching technologies, and (c) dial-up, Point-to-Point Protocol (PPP) and Integrated Services Digital Network (ISDN) circuit switching technologies. It must be reemphasized that the individual links within the WAN backbone are point-to-point links that use a multiplexed packet-switching architecture. It should be noted that point-to-point leased lines are typically only implemented over the last mile to the customer premises equipment and use packet-switching or circuit-switching on the WAN backbone. Lastly, all of these technologies map to the OSI physical and data link layers and are presented in detail below
WAN Multiplexing
Recall our previous multiplexing definition in OS as the ability to create multiple logical resources from a single physical resource. Consistent with this definition, in communications multiplexing refers to the ability to transmit several signals over the same channel simultaneously. With respect to WANS and packet switching, multiplexing is facilitated through encapsulation by placing identifying control information in the protocol data unit’s (PDU) headers. At the destination, the encapsulated control information is used to identify and demultiplex the transmissions . It must be noted that multiplexing takes place at almost all layers of TCP/IP hierarchy. As presented earlier, TCP and UDP multiplex multiple application layer conversations through the use of port numbers. Both TCP and UDP PDU headers contain specific application addresses encoded as 16-bit source and destination ports that provide session multiplexing from a single computer or IP address. As introduced earlier, the IP layer receives both UDP and TCP PDUs from the transport layer and transmits encapsulated PDUs as packets, As introduced earlier, both wireless and fiber optic technologies utilize frequency and wave-length multiplexing at layer one.
With respect to packet switching, multiplexing allows a single physical communications channel to be efficiently shared by multiple logical communications. A WAN’s internal links utilize statistical time-division multiplexing that transmit frames serially but accommodate many logical connections to co-exist on a single physical line. With this basis it is intuitive the multiplexing is critical to the WANs serial point-to-point architecture. In accord with this paper’s topic, the discussion of multiplexing will be constrained to the statistical time-division multiplexing used in conjunction with packet switching. As identified above, it must be reemphasized that WANs operate at layers one and two and therefore forward frames. This terminology can cause confusion when discussing packet switching since the word packet is often used to describe layer three PDUs. Consistent with the literature review, the word packet will be used generically and interchangeably with frames.
WAN Devices
As a basis, WANs contain the following devices that map to OSI layers one and two: (a) modems and digital service units/channel service units (DSU/CSU) that provide signal translation to the specific WAN communications medium, (b) routers that provide link connectivity and WAN interface ports, (c) communication servers that serve as concentrators for WAN access, and (d) WAN networking devices such as the PSNs introduced above. Modems and DSU/CSUs are layer one devices that convert an end device’s or router’s signal suitable for transmission over a communications medium. A modem converts digital signals to analog signals suitable for the particular communications medium and vice versa. A DSU/CSU performs similar signal translation between a router and its digital serial lines. As introduced in previous coursework, a router is a layer three device whose primary responsibility is to forward IP packets to local and remote networks. Note that in a WAN, routers primarily operate at layers one and two. WAN connections to the router can be separate or integrated in a single CSU/DSU . Examples of communication servers include the devices that connect and aggregate end-users to a WAN (e.g. Internet) through cable and digital subscriber lines (DSL). Lastly, WAN networking devices include PSNs introduced above and examples of PSNs include ATM switches, Frame Relay switches and plain old telephone service (POTS) switches.
For subsequent analysis it is also important to distinguish between customer premises equipment (CPE), data communications equipment (DCE) and data terminal equipment (DTE). CPE may be either owned or leased from the service provider and connects to the network owned by the service provider . Typically, the customer is responsible for managing the CPE. A DTE is either a modem or a DSU/CSU that places data on local loop for transmission to the DCE . A DCE in turn connects to the service provider’s central office (CO). To put this in context with respect to a typical home DSL network Internet connection. A user may connect their PC to an all-in-one wireless access point (AP)/VOIP router. This router may be owned by the user or alternatively may be provided (e.g. leased) from the service provider. Note that it is the customer’s responsibility to securely configure this AP/router. This CPE router will be connected to a DSL modem. This modem will in most cases be provided by the service provider. The modem will in turn connect to the service provider’s DSL Access Multiplexor (DSLAM) at the other end of the local loop that in turn connects to the CO.
As previously introduced, communications can be either synchronous or asynchronous. It is well known that computers that rely on quartz timers cannot be synchronized. With this basis, synchronous communications require the transmission of a separate but associated clocking mechanism with each transmission. Intuitively it makes sense for the communications carriers DCEs to provide the clocking mechanism since this is entirely under their management. With this basis, the DCEs provide clocking and synchronization to the DTEs.
WAN Models and Protocols
As previously introduced, WANs are responsible for OSI layer one and layer two functionality. The OSI physical layer (i.e. layer one) provides specifications of the electrical, mechanical, operational and functional components of a communications service . The OSI data link layer (i.e. layer two) defines how data is encapsulated in frames and how these frames are transferred . Physical connections were presented earlier and will not be repeated here. It is worthwhile to discuss WAN layer two data link layer protocols since there is a significant difference between LAN and WAN layer two protocols. WAN data encapsulation methods correspond to the individual WAN technologies and communications equipment and include: (a) PPP, (b) High-Level Data Link Control (HDLC), (c) Frame Relay, (d) ATM, and (e) the emergent Multiprotocol Label Switching (MPLS). With this basis it is obvious there are a wide range of layer two WAN protocols. HDLC will be discussed in this section since it is primarily used within the WAN. The remaining protocols will be analyzed with their respective WAN technologies below however note that a complete analysis of WAN encapsulation methods is beyond the length constraints of this paper.
As introduced above, a WAN architecture is comprised of serial point-to-point links that connect networks and multiplex packet based transmissions. HDLC is an International Organization for Standardization (ISO) encapsulation standard however it is open for interpretation evidenced by the existence of numerous proprietary extensions in data field . To provide examples of HDLC’s multiple versions, consider that it provides specifications for synchronous or asynchronous, connectionless or connection-oriented, and point-to-point or point-to-multipoint communications. HDLC’s extensibility has resulted in numerous proprietary implementations. As a result of this extensibility, HDLC has become the default encapsulation type on Cisco synchronous serial interfaces however Cisco’s implementation is not interoperable. If multi-vendor equipment is deployed, the PPP standard is a more viable option .
It should be noted that WANs may alternatively be defined as the core layer in Cisco’s three layer hierarchical internetworking model . A complete analysis of Cisco’s three layer model is beyond the scope of this paper however this model may provide a more comprehensive perspective than the OSI model. The three layer model provides additional insight into network design since it also addresses a network’s business goals and financial constraints. When WANs are described with respect to the Cisco three layer model, the WAN is the backbone or core layer and includes the high speed switches, routers and cables that facilitate reliable low latency high speed transmissions. The Cisco core layer also includes the private branch exchange (PBX) and core layer multiplexors. The Cisco three layer model will be briefly discussed in the paper’s conclusion.
Point-to-Point Networks
A point-to-point network provides a pre-determined end-to-end communications path from CPE to a DCE, through the WAN to a remote DCE connected CPE. Note that a point-to-point network is also known as a serial connection or a leased line and these terms may be used interchangeably . As identified above, the DCE provides clocking and signal transformation to the carrier’s specific channel format (Cisco Systems Inc., 2004). The WAN and therefore the point-to-point topology may include a number of intermediary devices that use packet-switched multiplexing technologies within network. With this basis, the distinction between point-to-point networks and circuit-switching may be negligible citing that certain Cisco authors categorize leased lines as circuit-switched connections (Cisco Systems Inc., 2004). Also note that these leased line connections may only specify a point to point connection to the service provider and therefore the Internet rather than pre-determined end-to-end connectivity (Cisco Systems Inc., 2003; Cisco Systems Inc., 2004).
Leased lines provide several advantages with respect to other WAN technologies. Leased lines provide dedicated bandwidth with very little latency or jitter and constant network availability . This makes leased lines very attractive for mission critical and business transaction data. Leased lines require minimal expertise to install and maintain and offer a high Quality of Service (QoS). Leased line bandwidth is limited only by the physical nature of communications medium and the price users are willing to pay for the dedicated bandwidth .
Leased lines have several disadvantages when contrasted with other WAN technologies. First, leased lines are the most expensive type of WAN access. The cost of leased lines depends on the required bandwidth, QoS and distance of the connection. Second, an organization’s bandwidth usage and needs are intuitively variable. Leased lines provide fixed capacity that results in wasted bandwidth when network traffic is minimal. Leased lines offer limited flexibility since the carriers provide a fixed and limited network capacity. If an organization needs more bandwidth, they must contact the carrier and provision additional bandwidth. Examples of leased lines connections are T-Carriers (e.g. T1 and T3/E3) and the more recent broadband DSL and cable connections. A brief summary of these implementations follows.
T-Carriers, DSL and Cable Connections
T1 is an always on, high-speed, long haul digital network developed by American Telephone & Telegraph (AT&T) to provide 1.544 Mbps transmission capacity (Teare, 2004). T1 prices continue to fall and organizations that need more dedicated capacity can avail themselves to the higher capacity T3 or T4 leased lines (Cisco Systems Inc., 2003). DSL is an always-on connection using telephone lines to connect subscribers point-to-point to a carrier and therefore the Internet . DSL allows utilization of existing local loop telephone lines and therefore provides accessible access to many communities. As identified above, a DSL modem will be required to convert layer two Ethernet signals to a frame suitable for transmission to CO. This local line is able to multiplex telephone voice analog and data on the same line using frequency division multiplexing. Multiple DSL connections are further multiplexed through the use of a DSLAM that uses time division multiplexing . DSL can be further differentiated as asymmetric DSL (ADSL) that provides higher download bandwidth than upload or symmetric DSL (SDSL) that provides equal capacities in both directions. Cable is defined as IP over Ethernet (IPoE) Internet service. In this environment the cable companies use existing coaxial cable television cables to multiplex television and data communications services. These providers are able to offer simultaneous download of television and full duplex data communications.
Packet Switching
As introduced at the outset, a WAN may be characterized as a shared serial packet-switched multiplexed communications channel . It should be noted that this paradigm supports both packet switched or circuit switched communications (Cisco Systems Inc., 2004). Packet switching is a connectionless paradigm where there is no dedicated path between source and destination . In this model, data packets may travel different routes between source and destination and it is the responsibility of upper layers to provide reliability and reconstitution of messages . Packet switched transmissions are divided into small pieces (e.g. packets,) and are multiplexed onto high-capacity interconnections . Functionally, packet switching involves routing small messages between source and destination based on address carried within an encapsulated packet .
Packet switching offers several advantages when contrasted with other WAN technologies. The primary advantage of packet switching as established by ARPANET is that communications channel can be shared among computers and their applications . Recall that a WAN is comprised of serial links. A packet switched paradigm can reduce the number of required links in the network as any link can support multiplexed data and provide end-to-end transport . Packet switching provides more efficient use of network resources since multiplexed transmissions can be maximized to meet the individual link’s capacity. In contrast to leased lines and circuit switching, packet switching allows excess bandwidth to be allocated and used by other traffic. Recall that WANs are typically pay-for services and the use of packet switching allows users to pay for their bandwidth usage. With this basis, users only pay for the bandwidth they need and therefore do not waste bandwidth similar to point-to-point connections.
Due to packet switching’s simplicity, its connectionless model can support a wide range of Internet services . As an example packet switching provides the necessary foundation to create both permanent virtual circuits (PVC) and switched virtual circuits (SVC) detailed below in circuit switching. The primary disadvantage of packet switching is that a portion of the network’s capacity is necessarily consumed by its overhead . To illustrate packet switching technology further, Frame Relay and ATM will be explored below.
Frame Relay
From a historical context, Frame Relay was developed as the successor to the now obsolete X.25 (Cisco Systems Inc., 2004). Frame Relay supplanted X.25 since it operates at the physical and data link layers rather than at X.25’s operation that included the network layer (; Teare, 2004). Consistent with modern WAN technologies, this allows Frame Relay to perform its processing in hardware resulting in a far more streamlined protocol. Frame relay is also more efficient than X.25 since it does not perform error checking. This is facilitated by today’s more reliable communications mediums that are less prone to errors .
From an implementation standpoint, Frame Relay specifies connections between data terminal equipment (DTE) and data communications equipment (DCE) . Frame Relay was developed to provide a more cost-effective alternative to leased lines since circuits could be established as necessary through a single WAN interface. Frame Relay now operates on nearly any serial interface that includes ISDN or DSL lines (Cisco Systems Inc., 2004, ). With this basis, Frame Relay is cost-effective because LAN access can be provided without additional expense. Frame Relay provides high bandwidth low latency packet switching and supports both PVCs and SVCs through Data Link Connection Identifiers (DLCI). With this basis it is apparent Frame Relay is used over many different interfaces and continues to offer reliable connection services.
ATM
Asynchronous Transfer Mode (ATM) is connection oriented, cell-based networking technology used in both enterprise LAN backbones and WAN links. ATM is designed to take advantage of high-speed transmission media such as E3, SONET, and T3, or any transmission media up to 10 Gbps. ATM is similar to X.25 as they both use packet switching hardware to facilitate a connection oriented paradigm . ATM specifies 53 byte cell sizes (i.e. packet sizes) with a minimal five byte header and 48 octets of data . ATM’s cells may be considered to be fixed length packets that allow processing to occur in hardware reducing transit delays.
ATM’s high speeds make it widely applicable to multiple service types that include voice, video, or data . ATM’s small cell size is very similar to the TCP push functionality discussed in earlier. Recall TCP’s push functionality sent smaller sized segments to accommodate transmissions that were intolerant to delay. As a result of this functionality, ATM has become an international standard for multiple service types that include voice, video and data. It should be noted that while the ATM cell size has been specified as an invariable 53 bytes, in reality ATM can accept and deliver larger packets .
While ATM’s speed and capacity is attractive, its high speed efficiency also results in some drawbacks. To realize ATM’s efficiency, ATM requires specialized and expensive hardware and software that includes fiber optic cables for connections. Another disadvantage of ATM is that it is less efficient than protocols that use larger packets (e.g. Frame Relay and X.25). ATM’s smaller cell size results in 20% greater overhead when contrasted with these other technologies.
Circuit Switching
Some authors cite that circuit switching technology is interchangeable with connection oriented services in accord with the virtual circuit abstraction . Circuit switching can be further categorized as always-on PVCs or on-demand SVC as introduced above. In both cases, a connection is established between a source and destination. This connection oriented paradigm allows the end stations to communicate transparently as if they were attached through a permanent and dedicated hardware connection . Although a dedicated path is established, maintained and terminated through the WAN, it must be reemphasized that the intermediary route is most likely maintained through statistical time division multiplexing.
Connection oriented communications provide several advantages in contrast to other WAN technologies. Circuit switching can provide guaranteed capacity as no other entity can decrease the capacity of the circuit . Circuit switching supports point-to-multipoint connections (i.e. multicast) since it allows multiple sites to connect to a single source (e.g. video streaming). As identified in earlier, connection oriented communications are able to provide reliable transport services . Reliable transport service is a critical component to today’s mission critical transaction business communication. The primary disadvantage of circuit switching is that costs are fixed whether all or a portion of the bandwidth is used . While circuit switching provides dedicated bandwidth it must be noted that circuit maintenance requires administrative overhead that consumes effective throughput. To illustrate circuit switching further, the PPP implementation will be explored below.
PPP
Point-to-Point Protocol (PPP) provides point-to-point, router to router, host or router and host to host connections. PPP was developed as an interoperable and extensible mechanism for transporting IP traffic over high-speed links (Cisco Systems Inc., 2004). This interoperable flexibility has made it the most commonly used Internet WAN connection . PPP is defined in IETF RFC’s 1661 and 1332 and has been updated by RFC 2053 for specific vendor extensions. With this basis, PPP provides an interoperable standard for multivendor implementations. PPP provides its varied network connections over: (a) synchronous serial circuits, (b) asynchronous serial circuits, (c) basic rate interface (PRI) serial circuits, and (d) high-speed serial interface (HSSI) serial circuits . To provide definitional bases: (a) a synchronous circuit is typically a leased line, (b) an asynchronous circuit is dial-up connection, (c) a BRI is an ISDN connection, and (d) HSSI is used on high speed point-to-point links within the LAN. As cited above, PPP provides an interoperable multivendor protocol in contrast to the many proprietary HDLC implementations.
PPP provided WAN technologies with an important evolutionary step as it established: (a) a standard for the assignment and management of IP addresses across networks, (b) multiprotocol encapsulation, (c) multiprotocol multiplexing, and (d) extensible configuration parameters . Examples of PPP’s multiprotocol support are evidenced by the development of PPP over Ethernet (PPPoE) and PPP over ATM (PPPoA). As a connection oriented protocol, PPP must negotiate, establish and configure its connections. PPP’s extensible management is facilitated by PPP’s Link Control Protocol (LCP) and Network Control Protocols (NCPs). LCP and NCP allow PPP to establish, configure and test connections and different layer protocols. LCP and NCP may also provide authentication (e.g. PAP and CHAP) and link quality determination . Lastly LCP is responsible for link termination
Summary
WAN relevance and importance cannot be understated citing that the emergence and evolution of the Internet has changed business models, global economies, societies and even cultures (Friedman, 2005; Kotler & Keller. 2007; Laudon & Laudon, 2004; Robbins & Judge, 2007). Today’s business models are characterized by organic virtual teams that often cross geographic boundaries (Robbins & Judge, 2007). Implicit in this functionality is the need to provide remote access to centralized resources (Cisco Systems Inc., 2004). Organizations must also continually reach out to new constituencies and markets to remain competitive (Kotler & Keller, 2007). With this basis, organizations must understand the nuances of the Internet and WWW to facilitate robust secure information access and communications (Cisco Systems Inc., 2004).
While WANs can be described in relation to the OSI networking model, an increasingly attractive analysis can be performed based on Cisco’s three tier model, comprised of the core, access and distribution layers. In this model, the WAN may be defined as the core layer and is responsible for fast and reliable transportation of data across a network (Teare, 2004). Core networks enable fast reliable communications through load sharing across multiple redundant paths (Teare, 2004). The core layer may also be described as the backbone or foundation network because all other layers rely upon it (Cisco Systems Inc., 2003). With this basis, it is obvious that WAN’s are the backbone of today’s global economy. It is also obvious the Cisco three layer core’s description is not only in accord with WAN definitions but continues to define WAN technology with respect to business and communication needs. In conclusion, the Cisco three tier model’s definition may be more relevant than the OSI model as it accommodates the business functionality and perspective necessary to remain competitive.
References
Cannon, K., Caudle, K., & Chiarella, A. (2009). CCNA: Guide to Cisco networking (4th ed.). Boston, MA: Course Technology
Cisco Systems Inc. (2003). Internetworking technologies handbook (4th ed). Indianapolis IN: Cisco Press
Cisco Systems Inc. (2004). Cisco WAN technologies and components. Retrieved May 24, 2009 from http://www.ciscopress.com/content/images/1587051486/samplechapter/1587051486content.pdf
Comer, D. E. (2000). Internetworking with TCP/IP: Principles, protocols and architectures, (4th ed.). Upper Saddle River, NY: Pearson Publishing.
Graziani, R., & Johnson, A. (2008). Routing protocols and concepts: CCNA exploration companion guide. Indianapolis, IN: Cisco Press ISO 3309:1991. (2009). ISO/IEC 3309:1993. Retrieved May 24, 2009, from http://www.iso.org/iso/iso_catalogue/catalogue_ics/catalogue_detail_ics.htm?csnumber=855
Friedman, T. L. (2005). The world is flat: A brief history of the 21st century. NY: Farrar, Straus and Giroux
Kotler, P. & Keller, K. L. (2007). Marketing management (12th ed.). Upper Saddle River, NJ: Pearson Publishing.
Laudon, K. C., & Laudon, J. P. (2004). Management information systems (8th ed.). Upper Saddle River, NY: Pearson Publishing.
McQuerry, S. (2008). CCNA Self-Study: Introduction to Cisco networking technologies Part 1 (ICND1) 640-822 (2nd ed.). Indianapolis, IN: Cisco Press
multiplex. (2009). In Merriam-Webster Online Dictionary. Retrieved May 14, 2009, from http://www.merriam-webster.com/dictionary/multiplex
Robbins, S. P., & Judge, T. A. (2007). Organizational Behavior. Upper Saddle River, NJ USA: Prentice Hall
Rosen, E., Viswanathan, A., & Callon, R. (2001). RFC 3031: Multiprotocol label switching architecture. Retrieved May 23, 2009, from http://www.ietf.org/rfc/rfc3031.txt
Teare, D. (Ed.). (2004). CCDA self-study: Designing for Cisco internetwork solutions (DESGN). Indianapolis IN: Cisco Press