Home Lab Archive - Part 2

Tuesday, 21 March 2000 19:00 administrator
Print

Set Up Your Home Router Lab! Part 2 


by Erik Westgard and Michelle Truman 

March 2000 

Resources You Need
Various CCNA exam textbooks have information on cabling and introductory router setup. To supplement that, we use the Cisco Web site at http://cio.cisco.com, the Cisco documentation CD, or one of several router configuration reference books for basic hardware information. Two books we have found useful includeIntroduction to Cisco Router Configuration by Laura Chappell and Cisco Router Configuration by Allan Leinwand, Bruce Pinsky, and Mark Culpepper. Be sure to read reviews and ask around before you buy to get good quality study and reference material. Some of the online book-buying sites even support book rating systems.

For direct connections there are two serial encapsulation options: PPP and HDLC. It's possible to buy a single "back to back" serial cable with one DCE and one DTE end, or you may wish to buy both a DCE and DTE cable and connect them back to back. See Part 1 of our series for a rundown on cabling.

Tip: Note that using a router as a frame relay switch means you have devoted the serial interfaces to that purpose, so it's good to have four routers in that case. We'll devote a future article to building your own frame relay switch for advanced lab exercises. In Figure 1 there are two networks: Router B to A and Router A to C.

 


 


Figure 1. Back-to-back serial connections using encapsulation PPP or HDLC.

In Figure 2 there are also two networks represented from Router B to Router A and Router A to Router C. Note that Router A must tell router C how to reach Router B. This requires a routing protocol or static routes to be inserted on router A.


Figure 2. Back-to-back LAN connections, which can be Ethernet, using a hub or a cross-over cable, or token ring.

In Figure 3 we use Router A as a frame relay switch. If router A has only 2 serial ports, we can only use it to build a PVC (Permanent Virtual Circuit) from router B to router C. This means we can create only a single routed connection and makes it difficult to practice routing protocols. While it is possible to use subinterfaces on each router and create multiple PVC's between router B and C, this is still a directly connected network and does not give much practice with routing. If you are going to use a frame relay switch it is best to get something like a Cisco 4000 with 4 serial interfaces for multiple connections or to stick with back to back cabling and other layer 2 encapsulations.


Figure 3. Router A acts as a Frame Relay switch. Router B connects to Router C via one PVC, which is also one network.

Along with a target configuration, you should do a lab IP address plan. Additional planning could include IPX addresses, frame relay DLCIs, AppleTalk cable ranges, or Decnet node numbers and all layer 2 node and layer 3 network layer addresses you'll need for your lab. We suggest using RFC1918 private addresses, just to save yourself grief if you were to connect your network to the Internet or a production customer network. These private addresses aren't routable on the Internet, so they provide some measure of protection.

 

IP subnetting-the practice of dividing IP network blocks into smaller subnets-is an extremely important skill when attempting any level of Cisco certification. It's also critical to get some practice with different ranges of addresses, so you can build a comfort level in working with classical A, B, and C addresses as well as the concept of CIDR (classless inter-domain routing). The non-Internet routable addresses available for you to use are as follows:

Net block range Pre-CIDR class designation CIDR block
10.0.0.0 - 10.255.255.255 (1 Class A Network) /8
172.16.0.0 - 172.31.255.255 (16 contiguous Class B networks) /12
192.168.0.0 - 192.168.255.255 (256 contiguous Class C networks) /16

 

Consider using a couple of different ranges to get some practice with different net block sizes. You may want to use several /24s from the 192.168.0.0 range for your serial interfaces. You could use the 10.0.0.0 network for all of your LAN addresses and use 172.16.0.0 networks for loopback addresses and other miscellaneous tasks.

Resources
To read more about the concept of CIDR and classless addressing check out the following references:
  • For coverage of CIDR, visit:http://www.ietf.org/rfc/rfc1519.txt

  • For private addressing, visithttp://www.ietf.org/rfc/rfc1918.txt
  •  

    One of the easiest beginner pitfalls is to get the subnet masks wrong in a lab. If you settle on using Class A 10.0.0.0 network subnetted into /24 blocks, be sure you're consistent with the masks for every device on a particular network. One of us (Erik, to be specific!) can remember lots of grief from setting up a /16 mask on the TFTP server PC, and /24 masks on the router interfaces. It's possible to use differently sized masks for the same major network on different physical networks. But for beginners, just play it safe and keep all masks the same.

    You'll need console devices for direct access to the router interface. A PC makes a fine console, and even an older 486 machine or laptop can handle this job. The 2501 console kits come with a set of adapters, but I use the DB9, which fits the standard DB9 PC serial port that was first introduced with IBM PC/AT. Most of the other routers you may have in a lab use a DB25 console connection; but the little DB9 to RS232 adapter cables sold in any computer store work fine here. You can, of course, use straight through DB25 cables to the DB25 ports. Set your PC communications software (the standard terminal emulator feature included with Microsoft Windows works fine) to 9600 bps, which is the default on a Cisco router console interface, unless it's been changed in the router boot register.

    While setting up your console machine, you might want to make sure it has a TCP/IP stack and LAN card. You can then use that machine as your TFTP server. CCO has a free downloadable TFTP package (the 1.3M file TFTPServer1-1-980730.exe located at http://www.cisco.com/pcgi-bin/tablebuild.pl/tftpM/), and there are some other freeware versions. This will allow you to store IOS loads, and router configuration files on the PC hard disk. You can connect to the router via a 10BaseT crossover cable or a hub. One hint: If you're mixing older 10BaseT routers, and newer 100M PC LAN cards, make sure you have an autosensing hub; some 10/100 non-autosensing hubs won't translate the LAN adapter speeds. One last note: On the 4000, the default is to power the AUI port, not the 10BaseT port.

    On the subject of cabling, we suggest using only new 10BaseT cables. One of us (Erik, again!) had some old, used eight-wire telephone cables, with the correct RJ45 jacks, which seemed to work. These weren't rated for LAN speeds, and led me to suspect my 3104's Ethernet port. After an embarrassing trip back to the dealer for a "no trouble found" call, I tossed these out and bought new cables. If you look in the right places, these are around $5. There's a part of the spec for 10BaseT that suggests cables should be at least eight feet long.

    You may find your used router arrives at your door with a password set. Fortunately, if you search on CCO or even the comp.dcom.Cisco archives on deja.com, you can find the directions for "cracking" passwords. You're actually entering the break key from the console during boot and telling the router to ignore the configuration file, which has the passwords. It's most helpful to learn about the different settings of the boot register and to try booting the router from the subset IOS (usually) stored in ROM, vs. flash. Netbooting (loading IOS off of a TFTP server) is also mandatory if you have a really old router that needs to run a newer IOS in RAM or are doing any memory or IOS upgrades. It's also possible (although slow and tricky) to load IOS via the console or AUX ports via Xmodem if you're using Cisco 2600 or 3600 series routers.

    On the subject of IOS, used routers will normally have an older IOS loaded, like 9.1 or 10.1. If you have a SmartNet contract on a supported router, you can go right to the software center on CCO and download the version you need. Remember: You only get what you paid for in terms of feature sets. If your router came licensed for Enterprise, you can download a newer version of Enterprise from CCO. If it only came with IP and you want Enterprise (or IP Plus or Firewall), you need to purchase the appropriate upgrade package from Cisco or a dealer. The 2500s have feature packages you can buy that have a CD included with the IOS and a software loader. The whole licensing issue on out-of-support routers is tricky. On his 3000s Erik's running a version of 11.1 IP that was provided by Cisco's Y2K desk, since they say these models are Y2K-compliant on that release as a minimum.

    A brand new router, or one where a configuration file is invalid or unavailable, will prompt you to enter setup mode. This is a dialog-driven way to assign your router a name and some passwords and to get the basic interfaces up and running. This is where you should follow your addressing plan (and watch out for those mask issues). You can also enable some of the routing protocols from the setup dialog. If you choose to perform configurations manually (a good idea for certification practice anyway) just say, "no," when asked to save the configurations from setup mode.

    Ping is your friend here. You can ping the routers from each other and from your TFTP server at the MS-DOS prompt. The Cisco Discovery Protocol -- a media and protocol independent way for Cisco routers to find each other on any network -- is available on newer versions of IOS. If your IP or IPX addressing isn't right but the routers are cabled properly, (and you have the clocking set properly on the serial ports), CDP will see them. Another valuable networking tool available to you from the router prompt is the traceroute command. You can use traceroute to verify the path taken from the host you're on to any given destination.

    Once you have two or more routers communicating, you'll want to enable some form of IP routing. The old, standard routing protocol and perhaps the simplest to use is to enable Router RIP (Routing Information Protocol). You merely enable the protocol and type in the network you wish to route, and RIP routes will begin to flow. Your directly connected interfaces won't show up in the RIP table though. You may wish to use loopback interfaces-virtual IP networks-on each router to simulate routing traffic.

    Other routing protocols you should play with are OSPF, IGRP, and EIGRP. The latter two are proprietary to Cisco, while OSPF and RIP are RFC-developed protocols. Each routing protocol has different advantages and disadvantages. A challenge for you is to develop a good understanding of how a distance vector routing protocol (RIP) differs from a Link State Protocol, which differs from a hybrid (EIGRP), and finally a path vector routing protocol (BGP).

    This should be enough to keep you very busy until our next article, in which we'll discuss the details of building a frame relay switch, distributing between two different routing protocols, and dealing with "routed" protocols such as IPX, AppleTalk, and SNA.