Point-to-Point Protocol (PPP)

Let’s spend a little time on Point-to-Point Protocol (PPP). Remember that it’s a Data Link layer protocol that can be used over either asynchronous serial (dial-up) or synchronous serial (ISDN) media. It uses Link Control Protocol (LCP) to build and maintain data-link connections. Network Control Protocol (NCP) is used to allow multiple Network layer protocols (routed protocols) to be used on a point-to-point connection.

Since HDLC is the default serial encapsulation on Cisco serial links and it works great, why and when would you choose to use PPP? Well, the basic purpose of PPP is to transport layer 3 packets across a Data Link layer point-to-point link, and it’s nonproprietary. So unless you have all Cisco routers, you need PPP on your serial interfaces—the HDLC encapsulation is Cisco proprietary, remember? Plus, since PPP can encapsulate several layer 3 routed protocols and provide authentication, dynamic addressing, and callback, PPP could be the best encapsulation solution for you instead of HDLC.

Figure 1 shows the protocol stack compared to the OSI reference model.

PPP contains four main components:

EIA/TIA-232-C, V.24, V.35, and ISDN A Physical layer international standard for serial communication.

HDLC A method for encapsulating datagrams over serial links.

LCP A method of establishing, configuring, maintaining, and terminating the point-to-point connection.

NCP A method of establishing and configuring different Network layer protocols. NCP is designed to allow the simultaneous use of multiple Network layer protocols. Some examples of protocols here are IPCP (Internet Protocol Control Protocol) and IPXCP (Internetwork Packet Exchange Control Protocol).

Burn it into your mind that the PPP protocol stack is specified at the Physical and Data Link layers only. NCP is used to allow communication of multiple Network layer protocols by encapsulating the protocols across a PPP data link.

High-Level Data-Link Control (HDLC) Protocol

The High-Level Data-Link Control (HDLC) protocol is a popular ISO-standard, bit-oriented, Data Link layer protocol. It specifies an encapsulation method for data on synchronous serial data links using frame characters and checksums. HDLC is a point-to-point protocol used on leased lines. No authentication can be used with HDLC.

In byte-oriented protocols, control information is encoded using entire bytes. On the otherhand, bit-oriented protocols use single bits to represent the control information. Some common bit-oriented protocols include SDLC, LLC, HDLC, TCP, and IP.

HDLC is the default encapsulation used by Cisco routers over synchronous serial links. And Cisco’s HDLC is proprietary—it won’t communicate with any other vendor’s HDLC implementation. But don’t give Cisco grief for it—everyone’s HDLC implementation is proprietary. Figure 1 shows the Cisco HDLC format.

 As shown in the figure, the reason that every vendor has a proprietary HDLC encapsulation method is that each vendor has a different way for the HDLC protocol to encapsulate multiple Network layer protocols. If the vendors didn’t have a way for HDLC to communicate the different layer 3 protocols, then HDLC would only be able to carry one protocol. This proprietary header is placed in the data field of the HDLC encapsulation.

So let’s say you only have one Cisco router, and you need to connect to a non-Cisco router because your other Cisco router is on order. What would you do? You couldn’t use the default HDLC serial encapsulation because it wouldn’t work. Instead, you would use something like PPP, an ISO-standard way of identifying the upper-layer protocols. You can check out RFC 1661 for more information on the origins and standards of PPP. Let’s discuss PPP in more detail and how to connect to routers using the PPP encapsulation

Data Terminal Equipment and Data Communication Equipment

By default, router interfaces are data terminal equipment (DTE), and they connect into data communication equipment (DCE) like a channel service unit/data service unit (CSU/DSU).

The CSU/DSU then plugs into a demarcation location (demarc) and is the service provider’s last responsibility. Most of the time, the demarc is a jack that has an RJ-45 (8-pin modular) female connector located in a telecommunications closet. Actually, you may already have heard of demarcs.

If you’ve ever had the glorious experience of reporting a problem to your service provider, they’ll usually tell you everything tests out fine up to the demarc, so the problem must be the CPE, or customer premises equipment. In other words, it’s your problem not theirs.

Figure 1 shows a typical DTE-DCE-DTE connection and the devices used in the network. The idea behind a WAN is to be able to connect two DTE networks through a DCE network. The DCE network includes the CSU/DSU, through the provider’s wiring and switches, all the way to the CSU/DSU at the other end.

The network’s DCE device (CSU/DSU) provides clocking to the DTE-connected interface (the router’s serial interface). As mentioned, the DCE network provides clocking to the router; this is the CSU/DSU.

If you have a nonproduction network and you’re using a WAN crossover type of cable and do not have a CSU/DSU, then you need to provide clocking on the DCE end of the cable by using the clock rate command.

Point-to-Point Protocol over Ethernet (PPPoE)

Used with ADSL services, PPPoE (Point-to-Point Protocol over Ethernet) encapsulates PPP frames in Ethernet frames and uses common PPP features like authentication, encryption, and compression. But as I said earlier, it’s trouble if you’ve got a badly configured firewall. This is a tunneling protocol that layers IP and other protocols that run over PPP with the attributes of a PPP link so they can then be used to contact other Ethernet devices and initiate a point-to-point connection to transport IP packets.

Figure 1 displays typical usage of PPPoE over ADSL. As you can see, a PPP session is connected from the PC of the end user to the router and the subscriber PC IP address is assigned by the router via IPCP.

Figure 1 PPPoE with ADSL

PPPoE is used to equip custom PPP-based software with the ability to deal with a connection that’s not using a serial line and to be at home in a packet-oriented network environment like Ethernet and to allow for a custom connection with login and password for Internet connection accounting. Another factor is that the opposite side of the link’s IP address is only given to it and available for the specific period that the PPPoE connection is open, so reusing IP addresses dynamically is permitted.

PPPoE has a discovery stage and a PPP session stage (see RFC 2516) that works like this: First, a host begins a PPPoE session, during which it has to execute a discovery process so it can determine the best server to meet the needs of the client machine’s request. After that, it has to discover the Ethernet MAC address of the peer device and create a PPPoE session ID. So even though PPP delimits a peer-to-peer relationship, the discovery part is innately a client-server relationship.

ADSL

ADSL supports both voice and data at the same time, but it was created to allot more bandwidth downstream than upstream because it’s best for residential subscribers that usually need more downstream bandwidth for doing things like downloading video, movies, and music; online gaming; surfing; and getting emails—some that include sizeable attachments. ADSL will give you a downstream rate from 256Kbps to 8Mbps, but anything going upstream is only going to reach around 1Mbps.

POTS provides a channel for analog voice transmission and can transmit without a problem with ADSL over the same twisted-pair telephone line. Actually, depending on the type of ADSL, not just two, but three information channels commonly utilize the same wiring at the same time. This is why people can use a phone line and an ADSL connection at the same time and not affect either service.

ATM is the Data Link layer protocol typically used over the DSL layer 1 connection from the CPE that’s terminated at what’s known as the DSLAM—an ATM switch that contains DSL interface cards, or ATU-Cs. After ADSL connections meet their end at the DSLAM, it switches the data over an ATM network to something called an aggregation router—a layer 3 device where the subscriber’s IP connection then expires.

You know by now how important encapsulation is, so as you’ve probably guessed, any IP packets over an ATM and DSL connection must have this done. This happens in one of three ways (PPPoE,RFC1483 Routing,PPPoA), depending on your interface type and the service provider’s switch.

Digital Subscriber Line (DSL)

Coming in second in our subscriber-based popularity contest is DSL (digital subscriber line), a technology that uses your garden-variety copper phone wires to give you high-speed data transmission. DSL requires a phone line, a DSL modem (often included with service), either an Ethernet card or a router that has an Ethernet connection, and someone that can provide service wherever you happen to be located.

The acronym DSL originally meant digital subscriber loop, but now its meaning has morphed to digital subscriber line. DSL group types fall into two categories based upon the upstream or downstream speed connections:  

Symmetrical DSL The speed for both downstream and upstream connections are equal, or symmetrical.  

Asymmetrical DSL Different transmission speeds occur between two ends of a network— downstream speed is usually faster.  

Figure 1 shows an average home user with xDSL, which is a transmission technology that moves data over copper pairs.

xDSL connection from home user to central office

 All types of DSL are layer 1 technologies.
ATU-R = ADSL Transmission Unit - Remote
ATU-C = ADSL Transmission Unit - Central

The term xDSL covers a number of DSL variations, such as ADSL, high-bit-rate DSL(HDSL), Rate Adaptive DSL (RADSL), Synchronous DSL (SDSL), ISDN DSL (IDSL), and very-high-data-rate DSL (VDSL).

DSL flavors that don’t use the voice frequencies band, like ADSL and VDSL, allow DSL lines to carry both data and voice signals simultaneously. Others, like SDSL and IDSL, that occupy the complete frequency range, can only carry data. And by the way, the data service that the DSL connection gives you is always on.

The speed that DSL service can offer depends on how far you are from the CO—the closer the better. In fact, you can blaze at rates up to around 6.1Mbps if you’re physically close enough!

Differences between DSL and Cable Modem Networking

Okay, before I talk about the serial encapsulation connections used on Cisco routers (HDLC, PPP, and Frame Relay), I’m going to discuss cable modems and DSL (including ADSL and PPPoE) as solutions for connections to wide area networks because I think it will really help you understand the practical differences between DSL and cable modem networking.

DSL and cable Internet services truly do have a lot in common, but they still have some basic, essential differences that are important for you to understand:  

Speed Most would say that cable is faster than DSL Internet, but cable doesn’t always win the race in real-world use.  

Security DSL and cable are based on different network security models, and until recently, cable has been the reputed loser in this contest. But now, it’s pretty much a toss-up, and both offer adequate security that meets the needs of most users. And when I say adequate, I mean that there are still some very real security issues relating to both alternatives, no matter what your ISP says!

Popularity Cable Internet is definitely “best in show” in the U.S., but DSL is beginning to catch up.  

Customer Satisfaction Here, the reverse is true—in the U.S., DSL is top dog. But still, do you know anyone that’s really totally satisfied with their ISP?

Figure 1 shows how a connection can terminate from modems to either a PC directly or a router. Typically, your router would run DHCP on that interface, as well as PPPoE. Both DSL and cable high-speed Internet services are available to millions of residential and business consumers worldwide, but in some areas, only one (sometimes neither) service is available.

Figure 1 : Broadband access using cable or DSL

                                                     Always-on Voice, Video, and Data Services
Surprisingly, some of the differences between DSL and cable modem have nothing to do with the actual technologies—it comes down to the individual ISP. All other things being equal, issues like cost, reliability, and quality of customer support for both installation and maintenance issues vary significantly from one provider to the next.  

Cable Cable is a great cost-effective connection for a small office or home office, or SOHO—yes, there is an acronym for everything! And even in larger organizations, cable (or DSL for that matter) can be great to have as a backup link.  

Here are a few cable network terms:

Headend This is where all cable signals are received, processed, and formatted. The signals are then transmitted over the distribution network from the headend.

Distribution network These are relatively small service areas that usually range in size from 100 to 2,000 customers. They’re typically composed of a mixed, fiber-coaxial, or HFC architecture, with optical fiber substituting for the distribution network’s trunk portion. The fiber forms both the connection from the headend and an optical node that changes light to radio frequency (RF) signals that are then distributed through a coaxial cable throughout the specific area being serviced.

DOCSIS (data over cable service interface specification) All cable modems and like devices have to measure up to this standard.  

Figure 2 shows where you would find the various types of networks and how the terms I just listed would be used in a network diagram.

Figure 2: Cable network and terms

The problem is that ISPs often use a fiber-optic network that extends from the cable operator’s master headend, sometimes even to regional headends, out to a neighborhood’s hubsite that then arrives at a fiber-optic node, which serves anywhere from 25 to 2,000 or more homes. (Don’t get me wrong, all links have problems—I’m not picking on cable—really!)

And here’s another issue: If you have cable, open your PC’s command prompt, and type ipconfig check out your subnet mask. It’s probably a /20 or /21 class B address. Oh my. You already know that’s either 4,094 or 2,046 hosts per cable network connection. Not good!

When we say “cable,” we really mean using coax (coaxial) cable for transmission. And CATV, or community antenna television, is now used as a means to offer cost-effective broadcasting to subscribers. Cable is able to provide voice and data, plus analog and digital video, without requiring you to pony up your whole paycheck.

Your average cable connection gives you a maximum download speed of 2Mbps. And remember—you have to share that bandwidth with all the other subscribers. As if that weren’t enough, there are other things like overloaded web servers and plain old Net congestion that factor in as well. But your email-checking neighbors really aren’t making that much of a difference. So who or what is? Well, if you’re an online gamer, you would likely notice a bit more lag during peak periods (which could be a matter of virtual life and death!). And if somebody in your neighborhood is uploading a large amount of data—like, well, an entire collection of pirated Star Wars movies—that could definitely max out the entire connection and bring everyone’s browser to a crawl.

Cable modem access may or may not be faster or easier to install than DSL, and your mileage will vary, depending on where you live plus a variety of other factors. But it’s usually more available and a tad less pricey, making it a winner by a nose. But no worries, if cable access isn’t available in your neighborhood, DSL is okay—anything is better than dial-up!