Configuring Cisco Routers for ISDN Paul Fischer $55.00 0-07-022073-5 |
![]() ![]() |
Reserve your copy at a Beta Bookstore near you! |
Contact Bet@books © 1998 The McGraw-Hill Companies, Inc. All rights reserved. Any use of this Beta Book is subject to the rules stated in the Terms of Use. |
External Concerns
The world of routers does not exist in a vacuum. Somewhere on your network, there will be a device not made by Cisco, or a wide area network connection provided by a third party. These can be classified as external concerns. They are as important to you as each Cisco router on your network; because without them, your routers are useless.
External concerns can be broken up into two types: WAN connections and hardware. WAN connections come in many different types including ISDN, Frame Relay, and point-to-point leased line. They are services provided by telecommunications companies for a fee. Hardware is anything connected to your network (but not made by Cisco) necessary for its function. This includes digital conversion devices like channel service units (CSU), data service units (DSU), and ISDN NT-1's. It also includes modems, external transceivers, hubs and switches (although we will not be dealing with them in this text.)
Telephone Companies
Every network administrator, even the accidental ones, can name at least five telecommunications companies fighting for their business. Each week, different arms of the telecom giants cold-call thousands of companies trying to increase their market share in local and long distance voice, frame relay, Internet, and leased line data services. For most network administrators this is an unwanted intrusion, but it does alert you as to which companies are competing for your business.
It is this competition that you must continually consider as you hang up on one rude sales person after another. Competition is good, since it drives prices down on your networking costs, whatever they may be. Although it is not feasible to change telecom companies on a regular basis, you can always use one phone company to drive down the price of the one you're using. If your prices on a line have been fixed for a while, you might consider having the competition provide you with a quote to replace the service. It's amazing how much money you can save.
Money should not be the only factor in choosing a telecommunications provider. Support, problem escalation, and reliability must also be considered. In the case of an ISP, you must also take into account the costs of changing IP addresses and domain name service (DNS) records. You might find that a few dollars a month doesn't make up for the headaches of converting your network. In addition, some leased lines have startup costs. If you're saving $50 a month but you have to pay a $1000 startup fee, then you would need to keep the line at lease 20 months just to break even.
The biggest problem overall, however, is the communications barrier between phone company people and network administrators. Each group has their own set of acronyms and phrases. They are used to dealing with people who speak their language. When the two groups meet, there is usually a lot of frustration due to insufficient vocabulary on both sides. Never doubt those phone company people want to get your lines working, just don't count on them being able to speak your language. Instead, you should learn some key phrases of their language to make sure you get what you need. Although you may not know exactly what it is you're asking for, Cisco has guidelines for almost all lines ordered from phone companies.
Telco History
Ordering lines from the Phone Company used to be a straightforward process. Before deregulation, there was only one source for data lines in a Local Access & Transport Area (LATA). A LATA is a geographic area where a specific telecommunications company used to have the exclusive rights to maintain and control the telephone infrastructure. In other words, there used to be only one source for local telephone and data lines. If you wanted to communicate, you had to do business with them.
This was not (and is not) true for connections between LATAs. If you wanted to connect between two or more areas, you had to use a long distance phone company known as an inter-exchange carrier (IXC). Many such companies exist to provide you service. Unfortunately, some may not target businesses your size, and may remain anonymous to you even though they provide excellent service at a fair price.
Ordering Lines from the Phone Company
When ordering a data line between multiple sites, you don't really need to know whether you are Inter-LATA or Intra-LATA. All you need to do is write up a request for a quote (RFQ) detailing your needs. You should include the name of the local contact, street address, and phone number. You should also ask for rate quotes for one through seven years of service.
The most important piece of information you provide are the phone numbers for each end of the connection. These will be used to determine what telephone company central office services your location, and therefore what the pricing will be. Telcos actually need only the area code and exchange (first three digits of a seven-digit phone number). However, they will usually not ask you for it in terms non-telco people would understand. Instead, put the acronym NPA/NXX in your personal linguistic computer (your brain) and translate that to something easier to understand. Area code and exchange will do, or the first six digits of a ten digit phone number.
Call all the telco companies that you are considering, and contact your sales person for the appropriate type of service you want installed (ISDN, frame relay, T1, etc), then fax your RFQ to them. This provides a paper trail and makes sure all of them have the same information. Make sure to place a deadline on responses. Two weeks is not outrageous. You may want to give them more time if more responses are important to you.
Undoubtedly, some of those asked to provide a quote won't respond. The good ones will let you know they either can't meet your deadline or don't provide the type of service you want. When you receive all the quotes, you will see they are all in different formats that make them hard to compare. This is not done by accident. This is done, in part, to confuse you a bit. Simply convert the quotes into a format that will let you perform an apples-to-apples comparison. Reduce the data by entering it into a spreadsheet, and calculate the cost for one through five years. Companies with large installation fees and smaller monthly payments may seem attractive, but you need to know where the break-even point will be in comparison to the others.
Once you choose a vendor, make sure you get as much information as possible. This includes phone numbers for their engineering department (which will install the line), and their internal order tracking number. They will commit to an installation date, but don't actually expect them to arrive on time, unless you keep on top of them. Do this by calling both the engineering department and the salesperson three or four days in advance. Make sure you are on their installation schedule and try to lock down a time (usually morning or afternoon) when they will be there. Make sure they have all your contact numbers so they can get a hold of you when they arrive.
No matter how you try not to answer the question, management always wants an estimate on when new lines will be installed. They don't realize the reason you don't want to answer the question is because they magically transform your estimate into a firm date. So, when things don't happen on that date, they are back at your door asking for an update. By planning for this in advance, you can avoid these problems. Take a page out of the Star Trek Engineers' Manual and add a week or two to what the phone company tells you. This has two benefits. It pads the schedule in your favor if the telco is late (they usually are.) It also makes you look like a miracle worker if it gets done early.
Sample RFQ
Here is a sample RFQ that was sent out to several telcos. Their responses have been summarized below. The data from each has also been reduced to let you see an apples-to-apples comparison. The information shown here is an example of their quote format and how to reduce the data to make comparison easier.
Dear Telco Salesperson:
Please provide a quote for a T1 leased line between the following locations:
Location: Newport News, VA Fairfax, VA.
NPA/NXX: 757-873 703-383
Address: 2020 Main St. 11225 Fender Dr.
Newport News, VA 23660 Fairfax, VA 22030
Contact: Laura Stein Alex Woo
Please provide pricing discounts for all long-term commitments you offer between one through seven years. This line will be a point-to-point data line with all 24 channels available.
Please fax a quote back to me within 2 weeks of receiving this RFP.
Below are three responses from AT&T, Sprint, and Worldcom. As you can see, the formats are extremely different, and not all the information is easy to pick out. This is partially due to the sales representative's familiarity with the author. Since they knew they were dealing with someone experienced in data line installations, they may not have been as concerned with format as they were with speedy delivery of the quote.
The first quote is from AT&T. You can see they provided pricing for a long-term contract (up to four years.) They are also willing to waive any installation charges if any sort of long-term contract is signed. Before taking a deal like this, make sure you know what the early termination fees are. It would be costly to sign a one-year contract and then have to owe them the installation fee, as well as the difference in price if the line were terminated early.
Sprint sent two quotes, one for a one-year term and the other for a two-year term. After follow-up phone calls, they agreed to send several more quotes for longer contract terms. They provided a lot of information, but the formatting made it difficult to determine exactly what costs were being charged. With a little perseverance, the costs became clear, and seemed very appealing.
WorldCom provided the most information, faxing out screen dumps from their order quoting system. They provided quotes for no contract, two, three, and five years. They also totaled the monthly recurring costs and the installation costs, which would be your first payment. The novice user might assume this is the total cost every month. This would make WorldCom's pricing look a higher than it actually is.
Below is the data reduction spreadsheet for all quotes received. The bids received were entered, and the total costs were computed. Whenever a vendor did not provide a quote, that vendor was omitted. This allows you to see only the quotes received for a particular contract length.
No Contract Pricing |
Install |
Monthly |
1-Year Cost |
AT&T |
$1,114.00 |
$ 6,266.00 |
$ 76,306.00 |
WorldCom |
$ 502.00 |
$ 4,373.09 |
$ 52,979.08 |
1-Year Pricing |
Install |
Monthly |
1-Year Cost |
AT&T |
$ - |
$ 5,048.00 |
$ 60,576.00 |
Sprint |
$ - |
$ 3,084.50 |
$ 37,014.00 |
2-Year Pricing |
Install |
Monthly |
2-Year Cost |
AT&T |
$ - |
$ 4,743.00 |
$113,832.00 |
Sprint |
$ - |
$ 2,869.50 |
$ 68,868.00 |
WorldCom |
$ 502.00 |
$ 4,219.42 |
$101,768.08 |
3-Year Pricing |
Install |
Monthly |
3-Year Cost |
AT&T |
$ - |
$ 4,438.00 |
$159,768.00 |
WorldCom |
$ 502.00 |
$ 4,118.85 |
$148,780.60 |
4-Year Pricing |
Install |
Monthly |
4-Year Cost |
AT&T |
$ - |
$ 4,367.00 |
$209,616.00 |
5-Year Pricing |
Install |
Monthly |
5-Year Cost |
WorldCom |
$ 502.00 |
$ 3,961.25 |
$238,177.00 |
There were only single quotes for four and five years so no direct comparisons could be done. However, if we move forward the last contract term pricing, we can fill in the blanks in the table. This allows better comparison based on the length of time you expect to be paying for the leased line.
No Contract Pricing |
Install |
Monthly |
1-Year Cost |
AT&T |
$1,114.00 |
$ 6,266.00 |
$ 76,306.00 |
Sprint |
no bid |
no bid |
no bid |
WorldCom |
$ 502.00 |
$ 4,373.09 |
$ 52,979.08 |
1-Year Pricing |
Install |
Monthly |
1-Year Cost |
AT&T |
$ - |
$ 5,048.00 |
$ 60,576.00 |
Sprint |
$ - |
$ 3,084.50 |
$ 37,014.00 |
WorldCom |
$ 502.00 |
$ 4,373.09 |
$ 52,979.08 |
2-Year Pricing |
Install |
Monthly |
2-Year Cost |
AT&T |
$ - |
$ 4,743.00 |
$113,832.00 |
Sprint |
$ - |
$ 2,869.50 |
$ 68,868.00 |
WorldCom |
$ 502.00 |
$ 4,219.42 |
$101,768.08 |
3-Year Pricing |
Install |
Monthly |
3-Year Cost |
AT&T |
$ - |
$ 4,438.00 |
$159,768.00 |
Sprint |
$ - |
$ 2,869.50 |
$103,302.00 |
WorldCom |
$ 502.00 |
$ 4,118.85 |
$148,780.60 |
4-Year Pricing |
Install |
Monthly |
4-Year Cost |
AT&T |
$ - |
$ 4,367.00 |
$209,616.00 |
Sprint |
$ - |
$ 2,869.50 |
$137,736.00 |
WorldCom |
$ 502.00 |
$ 4,118.85 |
$198,206.80 |
5-Year Pricing |
Install |
Monthly |
5-Year Cost |
AT&T |
$ - |
$ 4,367.00 |
$262,020.00 |
Sprint |
$ - |
$ 2,869.50 |
$172,170.00 |
WorldCom |
$ 502.00 |
$ 3,961.25 |
$238,177.00 |
You can see the price leader in this example is Sprint. The pricing difference is so great that almost all other concerns would be put aside in their favor. However, they may have particularly onerous early termination costs or other problems. Make sure you (or your lawyers) read and understand any contract they send you before you sign.
The important thing is to get the data yourself and make your own analysis. This is a particularly large distance for a leased line, and therefore the cost is extravagant. If you really needed to connect two sites as far away as these, you would be better off pricing out a frame relay connection. Frame relay is much less distance sensitive than a point-to-point leased line.
Factors in Choosing a Vendor
Business is never certain. You may believe that your new site will be around for years, but what if it isn't? If your break-even point is too far in the future, you will have cost your company money. You may think you can save money by choosing a longer time commitment because the monthly charges for longer contracts can be substantially cheaper. However, early termination fees are usually part of a long-term contract. If you choose a seven-year contract and the line is taken down in three years, you will owe the telco money.
The usual formula applied for early termination is the difference between the rate that you should have been paying and the rate you did pay times the number of months the line was active. For example, you order a line, which costs $500 for month five-year contract and $550 for a three-year contract. You sign a contract for five years, but cancel it in three. Under the standard formula you would owe the telco $50 (the difference between $550 and $500 a month) times 36 months (three years of service) or $1,800. This money will be due at the time of cancellation.
There are some extenuating circumstances concerning line termination fees. First, telecommunications companies don't always remember to go back and charge you. You shouldn't count on being overlooked, but neither should you remind your telco vendor to charge you for early termination. Second, you may be able to get the commitment on the line you are canceling switched over to another line on your network. This will delay you having to pay the early termination fee, although it will extend the commitment on the other line. At the very least, this tactic will delay the eventuality of paying the fee. In the interim, it can reduce the termination fee by extending the cost difference into a lower bracket. Make sure to ask about both early termination fees and moving commitments when you set up a line with a vendor.
Another factor in choosing a vendor is volume. Some vendors give you greater discounts based on the total monthly commitment you make with them. At a certain point you exceed the ability for other telecommunications companies to compete, unless they get a large chunk of your business. If your company does a lot of local and long distance business with a telco provider who offers volume discounts, you may find they are the least expensive.
Installation and Placement of Data Jack by Phone Company
The installation of your data line will almost never happen when the sales person says it will and will always take longer than expected. You can speed up the installation by making sure everything is in place before the telco installation engineer arrives. One way to attack the problem is from the destination backward. In this method, you determine exactly where you want the jack to be installed and then determine where the line needs to go to meet the telco lines in the building.
Start by determining where the router will be placed. If you can, install the equipment in advance and evaluate the space it is in. See if it can be accessed easily and if there is proper airflow for cooling. Once everything looks good, pick the location where you want the jack installed. Plan for the eventuality that you may not be there when the installation happens, and mark the spot on the wall, or tape a note indicating the location. This will help to eliminate confusion when the engineer arrives to do the installation.
Next, find out where the phone lines come into your office space. This is the area where the phone company's responsibility ends and yours begins. It is called the demarcation point, or in telco slang, simply the ‘demarc’. In office buildings the demarc may be a phone closet on each floor. In warehouse and smaller buildings, it may be a box on the wall or a sheet of plywood. When the time comes, you will need access to it. This access may have to come from a building engineer or someone else who manages the facility. Make sure you have multiple ways of finding this person, so that you can reach them when the telco installer finally arrives.
Finally, determine how you want the two points to be connected. The telco installer can usually put the jack anywhere you want. They will charge you time-and-materials for the work. They are usually skilled and efficient and you can usually blame wiring problems on the telco if there is a problem at a later time. The price for extending the jack to where you want it from the demarc is usually reasonable, and if you are only installing one line, it is a quick and inexpensive way to get the job done. If you are installing multiple lines from a demarc to a location (e.g., your computer room) you may want to contact a professional cable company and have them install a patch panel at each end to make multiple connections easier.
Extended Wiring Concerns
One thing to watch when running wiring from the demarc to your router is the path it takes. Electromagnetic interference (EMI) and radio frequency interference (RFI) sources between the demarc and the router can affect the performance of your data lines. You should instruct the cable installer (whether telco or independent) to steer clear of any electrical devices in the ceiling. This includes fans, air conditioning equipment, fluorescent light fixtures, and other sources of EMI and/or RFI.
If you are running more than a few data lines, you may want to talk to the telephone company about extending your demarc to where your routers will be. By doing this, you extend their responsibility right up to the jack on the wall next to your router. This can make for easier debugging of problems and eliminates finger pointing between the telco and independent cabling companies. Some telcos will try to limit the length of time the extension of the demarc is valid. Ask up front and raise a stink if they try to pull this on you after installation.
Post Installation Information
Before the telco installer leaves your site, make sure you have all information regarding the line. Each telco that has anything to do with the line will assign it a circuit identification number. Long distance and frame relay lines may have three or more circuit IDs. Dial-up lines will have associated phone numbers. ISDN lines will have System Profile IDs (SPIDs), as well as an ISDN switch type. Frame relay lines will have DLCI numbers. You should collect all this information into one place, and attach it to the copy of the installation work order left by the installer and the installer’s business card. If the business card does not have a name a phone number on it, make sure you write those down as well.
You should attach copies of any contracts relating to this line and the original quote to this cache of information, as well as the name and phone number of the sales person. Finally, get the phone numbers for reporting trouble on the line from all companies handling the line, and keep this within sight of the circuit ID numbers. With all this information in one place, you will have an easier time getting problems corrected on your lines.
Options on Carrier Formats
Not every data line is the same. Your challenge is to make sure the lines you order have their options properly set to mesh with your Cisco router. T1 and ISDN line configurations should be clearly conveyed to the telco to make sure they fit your needs. T1 lines have two general configuration options: B8ZS (Bipolar 8 Zero Substitution) / ESF (Extended Superframe Format), and AMI (Alternate Mark Inversion) / SF (Superframe Format). It is not important to know what they mean, as much as how they affect the T1 line.
A T1 line is a high capacity data circuit that contains 24 channels carrying either 56 or 64 Kbps in digital data or voice information. Although you can run voice or data over either configuration, you get the more bandwidth for data when configured for 64 Kbps per channel. B8ZS/ESF sets the channel speed to 64 Kbps, which gives you 1.536 Mbps of bandwidth. AMI/SF sets the channel speed to 56 Kbps, which gives you 1.344 Mbps by comparison. Typically, you want to use B8ZS/ESF for data and AMI/SF for voice. ISDN primary rate interface (PRI) lines, which run on T1 lines, must be set for B8ZS/ESF.
ISDN is a strange beast to configure. Since there are so many different types of telco central office switches for ISDN, there is a broad range of configurations. Some switches have few options, while others have hundreds on the same type of data line. To help you get past this problem, Cisco has provided a detailed list of what options they expect set on the telco ISDN switch in order for their routers to work. This list is ordered by switch type.
5ESS Custom |
BRI |
D |
2 B-channels for data Point to point Terminal Type = E 1 phone number assigned by telco MTERM = 1 Request delivery of Calling Line ID on Centrex lines May have to set speed of calls to 56 Kbps outside the local exchange |
D & V |
Only use this only with equipment supporting voice calls (i.e., 760 series, etc.) 2 B-channels for voice or data Multipoint Terminal Type = D 2 phone numbers, assigned by telco 2 SPIDs required MTERM = 2 Number of call appearances = 1 Display = No Ringing/Idle Call Appearances = Idle Auto-hold = No One touch = No Request delivery of Calling Line ID on Centrex lines May have to set speed of calls to 56 Kbps outside the local exchange.
|
||
5ESS NI1 |
BRI |
D & V |
Terminal Type = A 2 B-channels for voice and data 2 phone numbers, assigned by telco 2 SPIDs are required May have to set speed of calls to 56 Kbps outside the local exchange.
|
DMS-100 |
BRI |
D & V |
2 B-channels with both voice and data 2 phone numbers, assigned by telco 2 SPIDs Functional signaling Dynamic TEI assignment Maximum number of keys = 64 Release Key = No, or Key Number = No Ringing Indicator = No EKTS = No PVC = 1, for all BCS loads up to BCS 34, a PVC = 2 means NI1. This causes a problem because then a 2 digit TID is appended to the SPID. Use PVC=1 Request delivery of Calling Line ID on Centrex lines May have to set speed of calls to 56 Kbps outside the local exchange. Can have directory number 1 hunt to directory number 2 (it does cost a little extra) |
5ESS, 4ESS, and DMS-100 |
PRI |
D & V |
Line format = ESF Line coding = B8ZS Call type = 23 incoming channels and 23 outgoing channels Speed = 64 Kbps rate Call by Call capability 23B+D Trunk Selection Sequence = descending ( 23 ® 1) Set B+D glare = yield Only 1 phone number, assigned by telco May have to set speed of calls to 56 Kbps outside the local exchange. No SPIDs required |
Table 5-1. Specific ISDN line configuration information for a variety of ISDN switches.
Testing
When problems arise with WAN connections, your corner of the world will beat a path to your door. Many times the users start calling before the administrator even knows the line is down. Proper testing and troubleshooting methods will help you get your data lines back up quickly, with a limited amount of fuss. These methods stem from being able to break up the many components of your WAN connection into discrete parts and validating their operation one at a time.
Knowing how to test each piece is extremely important. There are many points of failure and without a proper debugging methodology, you can quickly become lost. One method of testing lines is from the inside out. To understand this method you must first understand how a data line is set up. In its simplest configuration, you have something like the figure below. In this figure, you have two Cisco routers connected to their own CSU/DSU. The CSU/DSU is then connected to the public switched telephone network (PSTN).
As you can see, the PSTN is the center of this connection. Because of this, the best place to start debugging is from the inside and move toward the router. Your telco provider is able to test your line from their network all the way to the equipment at your site. This proceeds (in order) from the smart jack, to your demarc, to your computer room jack, to your CSU/DSU, and finally, to your router.
The first piece of equipment on your site is the smart jack. It is a piece of telco equipment usually in or near your demarc and the telco can test it without any assistance from you. If the smart jack is bad, the telco will have to dispatch a repairperson. In this case, you should ask if they will need physical access to your site to make the repair.
The second piece the telco can test is the jack connecting your CSU/DSU to the demarc. You can only test this if you can put the jack into a hard loop back. In a hard loop back, all electrical signals sent by the telco are sent back to them. If you are setting up the line for the first time, you will want to make sure a RJ-45X jack is installed at the demarc and next to the CSU/DSU. A RJ-45X jack automatically goes into a hard loop back when the RJ-45 cable is removed from it. You can also create a loop back cable by cutting the head off one end of a RJ-45 cable and shorting pin 1 to pin 4 and pin 2 to pin 5.
Next, the telco can test all the way up to your CSU/DSU. They do this by trying to put the CSU/DSU into remote loop back. This creates a soft loop back through the CSU/DSU device, but only tests the CSU side. The final test you can perform is on the router serial port, router serial cable, and DSU side of the CSU/DSU. To test these, you need to go into global configuration mode on your router and set "loopback" on the serial port for the T1. This will allow the telco to pass data all the way to your router serial port and have it echoed back to them.
Regardless of the technology in use (ISDN, T1, Frame Relay), starting from the PSTN and working outward toward your router offers the best way to evaluate and solve wide area networking problems. The telco personnel can also run other tests besides loop back tests. One such test is the Bit Error Rate Test or BERT test. This can be useful in determining whether it is your equipment or theirs that is causing the problem.
Of course, it is always your job to blame their equipment; just as it is their job to blame your equipment. Instead of letting yourself get lead into a long round of finger pointing, try this when they say it's on your side. Tell them that you don't think enough testing and troubleshooting has been done, and you would like them to run another set of tests. Then make sure they verify that they are testing your line. It is always possible that through a simple human error, the telco engineer helping you is testing someone else's line. Hey, it happens! It happened to me!
Remember to always have the circuit ID and any other information about the line (like SPIDs) handy. This will allows the telco representative helping you to jump right on the problem, instead of waiting for you to provide vital information.
A Broader Testing Methodology
The basic starting point for advanced troubleshooting starts by answering the question, "What changed?" If a WAN line goes down and you haven't changed anything for weeks the best bet is to blame the telco. If you have been making changes, you should scrutinize them. In some cases, you may need to ask who has been in the room with the routers (Was it the guy whose watch always dies after a week?) or if work was done in the telco closet or near the demarc. Your problem could be as simple as someone bumping a piece of equipment, or tripping over a cable and damaging it.
A more likely cause is the telco. In fact, blaming the telco from the start and then investigating your own site may solve the problem more quickly. Changes are always being made at telco facilities that may affect your service for good or ill. Start by placing a call to them and then verifying your equipment. Even if they say it's not their fault, keep blaming them. The chance it is their fault is usually greater than it being yours.
After the problem is fixed, telco engineers like saying things like, "I looked at the line and I see no problem on it." Notice they say "no problem" in the present tense. The line goes down, you call, the line comes back up, and they see "no problem." They won't tell you they saw a problem and corrected it, and that brought the line back up. Seeing responses like that should make you wary of your line provider. It's OK to make mistakes once in a while, but it's not OK to lie about it or cover it up.
Whether your telco plays straight with you or not, you should always keep track of line outages. Record them on paper attached to your other information about the individual lines. This will allow you to review the line history after each incident. You should make sure you have all the trouble ticket numbers opened for your case and the explanations of the problem when they are closed. If you are dealing with long distance lines or frame relay, there may be several telcos carrying different parts of your line. Because of this, your telco may open trouble tickets with other telco companies to have them investigate the problem. If you have all the trouble ticket and contact numbers, you can inquire about the problem yourself.
The quickest way to a solution is to be a nuisance. Keep calling, stay on the line, and escalate to a supervisor if your line does not come up quick enough for you. The people at the telco end of the line probably have hundreds of lines to connect, disconnect, and fix. The only way you will become a priority is to be an annoyance that won’t go away. If the problem is a low priority, don't bug them about it. If the problem is a high priority, your only way to express this is to keep calling, stay on the line, and escalate the problem. Don't let them call you back, or you are at their mercy.
Of course, there are times when nothing you can do will get the line operational any faster. For example, if a major carrier looses a fiber optic cable bundle somewhere in middle America because Farmer John caught while plowing his field, hundreds or thousands of Mbps may be interrupted. If you have a single DS0 inside that bundle, chances are you will be among the last to be fixed. Simply put, the people who have larger connections (i.e., pay the most money) will get attended to first.
Keep them on the Line
When you are involving telco personnel in solving your problems, remember this nugget. It is corporate policy in many (if not most) of the telcos that they can't hang up on you. This means you can increase your priority and decrease down time, just by sitting on the phone. Usually they will put you on hold to look at something, and then come back to you. If you are not on hold, and they are going to call you back, you may get stuck on a pile with others, and they will get to you when they can. By hanging on the line, you are forcing the engineer at the other end to look at your problem. Remember, never hang up!
At time of installation
Be ready for the installation when it comes. If you are, you will certainly suffer fewer headaches than your peers. If all your equipment is in place (at both ends, if it is a point-to-point line), you can test the line while the engineer is still at your site. This will give you quick access to high level support, because the engineer will really want to leave to go to their next appointment. This is not your problem. If there is a problem when you install the line, you will want to keep the engineer there are long as possible to keep the lines of communication open.
Extended test period before the line is declared up for billing purposes
WAN lines are not just up or down. A line can be up, but experience high levels of data loss. This will cause poor performance with regard to your data throughput. Many times, these problems can first be detected at installation or shortly after. It is possible to ask the telco to place your line into testing mode for a short period. This is usually a day or two, during which you can perform high throughput data tests and check the lines for errors. In cases where the lines is bad, you can actually run the connection in an impaired state for weeks, until the telco fixes the problem. If the line is not officially declared operational, then you are not paying for it. If a problem exists at installation, make sure you don't pay for the line until it works at 100% capacity 100% of the time.
Understanding Telco Speak ( or: A guide to translating telephony speak into computer speak )
Telco people are just like you. They want to get problems fixed and help their customers. But sometimes it seems like they get frustrated dealing with non-telco people. This is mostly due to the language barrier that exists between telco people and mere mortals, that is the rest of us. The telco world has its own language, just like programmers, network administrators, and managers do. Someone not used to the insiders' language usually has a hard time relating requests and problems. To cut through this, learn some of the telco language, and tear down the language barrier. This will lead to quicker solutions to your problems.
Here are some telco language translations to make your life easier.
Telco |
Network Manager |
What is your line coding? What is your framing? |
Are you running B8ZS/EFS or AMI/SF? |
Can I loop your CSU remotely? |
Do you have remote loopback enabled in your CSU/DSU? |
I'll run a Bert test. |
I will run a Bit Error Rate Test to see if I can determine the problem. |
What is your NPA/NXX? |
What are the first six digits of your ten-digit phone number? |
Can I do intrusive testing? |
Can I take the line completely down to help determine what the problem is? |
Can you loop the remote CSU? |
If you have remote loopback enabled on the other end of your connection, can you loop that end back to you? |
Table 5-2. Telco slang.
Types of Lines
There are many different types of WAN lines. Each has its own particular setup and troubleshooting quirks, but the overall methodology stays the same. Here is a quick guide to some of the different types of WAN lines available.
T1
A T1 line is a high capacity digital leased line. Telco people sometimes call it a DS1. It consists of 24 channels, which can be configured for 56 Kbps, or 64 Kbps. Using all 24 channels for data is called a full T1 or a clear channel T1. You can also get a fractional T1, which denotes the use of anywhere from one to 23, channels. T1 lines are leased from telco companies and are only point-to-point. Each channel in a T1 is known as a DS0.
E1
An E1 line is a high capacity digital leased line. It is the rest of the world's equivalent to the T1 used in the United States. It consists of 30 channels, which can be configured for 56 Kbps, or 64 Kbps. Using all 30 channels for data is called a full E1 or a clear channel E1. You can also get a fractional E1, which denotes the use of anywhere from one to 29 channels. E1 lines are leased from telco companies and are only point-to-point. Each channel in a E1 is known as a DS0.
ISDN
ISDN lines come in two flavors: Basic Rate Interface (BRI) and Primary Rate Interface (PRI). They consist of some number of bearer (or B) channels, and one data (or D) channel. In the world of Cisco routers, the B channels carry all WAN data and the D channels carry call setup/tear down information.
Three types of calls can be placed on ISDN lines: 56 Kbps voice, 56 Kbps data, and 64 Kbps data. Obviously, all voice calls are made at 56 Kbps. There are two data options to support local and long distance calls. With long distance calls, the connection may not be ISDN end to end. When this happens, you will likely be using a single channel on a voice T1 line. Since voice T1 lines are setup AMI/SF, their 24 channels have only 56 Kbps of bandwidth. This is the time when you need to make 56 Kbps data calls. Local data calls are usually ISDN end to end, and therefore you have a full 64 Kbps available for data.
BRI
BRI lines are digital lines, usually procured through the local telco. Most are dial-up, but they can be ordered as leased lines. They consist of two 64 Kbps B channels and one 16 Kbps D channel. These can be used independent of each other or joined together for a 128 Kbps connection to a single site.
Older versions of Cisco IOS only supported the joining together two B channels for 128 Kbps of bandwidth. Starting with IOS version 11.2, Cisco now supports virtually unlimited joining together of channels, allowing you to combine up to 255 channels over multiple PRIs. Cisco only supports MPPP (Multi-link Point-to-Point Protocol) and not the ISDN standard of bonding. Cisco also does not support use of the D channel for anything other than call setup and tear down.
PRI
PRI lines are digital lines, which can be procured through a local or a long distance telco. PRI is an ISDN signaling standard, which is superimposed on a T1 or an E1. A T1 PRI consists of 23 B and 1 D channels. An E1 PRI consists of 29 B and 1 D channels. Because it runs on a B8ZS/ESF coded T1 or E1 line, all channels, B and D, are 64 Kbps. These can be used independent of each other or joined together for a higher bandwidth connection to a single site.
Currently, Cisco only supports the joining together two B channels for 128 Kbps of bandwidth. Cisco only supports MPPP (Multi-link Point-to-Point Protocol) and not the ISDN standard of bonding. Cisco also does not support use of the D channel for anything other than call setup and tear down.
56/64 Kbps
56 Kbps and 64 Kbps lines are also called DS0's by telco people. 56 Kbps lines can be point-to-point leased lines or dial-up, which are called switched 56 lines. 64 Kbps lines are only point-to-point leased lines. They offer lower cost connections between WAN sites while providing better reliability at a reasonable bandwidth.
Frame Relay
Frame relay is technically not a type of WAN line. The data travels not point-to-point, but rather through a frame relay provider's network. Connections within the network can change at any point in time, often without the customer noticing. Your sites lease a WAN line, be it 56/64 Kbps, ISDN, or T1, into the frame relay provider's local point of presence (POP). Since you only have to pay for a leased line to the frame relay POP (and not the whole distance between two sites), the costs for the leased lines is much cheaper.
Once inside the frame relay network, your data is routed between your sites based on your contract with the frame relay provider. You can have all your sites send their traffic back to a main location, or make a full or partial mesh between all the sites. You do not need to have all the data lines in your frame relay network be the same size. If you have several small sites, they may only require 56 Kbps or 64 Kbps connections to the frame relay cloud. Larger sites may get fractional T1 lines utilizing only two or three channels. The main site may have a full T1 to support all the traffic possible from all the remote sites.
Frame relay has the concept of committed information rates (CIR - pronounced "sear") and burstable information rates (BIR - pronounced "beer" [hic!]). The CIR is the minimum bandwidth that will be available for a particular site to communicate through the frame relay cloud. No matter how congested the providers network is, you will always have this minimum available for your networking needs. The BIR is the maximum bandwidth above your CIR that you can burst up to for short periods of time. Usually the CIR and BIR add up to the total port speed for your line. For example, if you have 128 Kbps fractional T1 connection into the frame relay cloud, you could have a 64 Kbps CIR and a 64 Kbps BIR. In this example you would always have a minimum bandwidth of 64 Kbps available to you, but when you need it you can burst up to 128 Kbps (assuming the network has the extra bandwidth to give you).
You could also setup a line with 0 Kbps CIR and 128 Kbps BIR, or the reverse. This will all depend on your needs and the rules of the frame relay provider. Some make you keep your CIR at a minimum of 50% of your port speed. Others may not care, but may have other rules regarding service such as only allocating CIR/BIR rates in 8 Kbps or 16 Kbps increments.
Some frame relay providers have additional services that you can purchase, such as network monitoring and customer adjustable CIR/BIR settings. The usefulness of network monitoring will depend on the individual customer and product offered by the provider. In some cases, a typical network management system (NMS) like HP Openview will give you the same information. So will some freeware products like MRTG, which graph router throughput and display it on a web page. For more information on MRTG, see the Appendix.
One nice thing about frame relay is its flexibility. It requires no customer changes to have the provider change the CIR/BIR ratio. This allows you to respond to user need for extra bandwidth by scheduling ratio changes with your provider. You should find out what the delay is from the time of your request to the time of implementation. This will let your users schedule changes to the network far enough in advance for you to implement them. In fact, some providers offer a service where you can go in on the fly and make changes to the CIR/BIR ratios. This usually costs a premium, but for larger sites it can save more than it costs.
A sample frame relay diagram is shown below. As you can see, there is one central site with a T1 local loop, two remote sites with T1 local loops, and two remote sites with 64 Kbps local loops. The main site has a fractional T1 line with a port speed of 256 Kbps. This is divided into 128 Kbps CIR and 128 Kbps BIR. All told, the other sites have 224 Kbps CIR, and 160 Kbps BIR. In this case, all sites communicate only with the central site.
Anyone with simple math skills can see the potential for disaster. Should all sites start transmitting at once, they can easily swamp the central site. But how often does this really happen? In truth, it will depend on your network users and applications, but rarely will the central site be flooded with traffic beyond its capacity. This principle is called "over selling bandwidth."
Because not all users on a network transmit or receive data at once, you do not need to provide 100% bandwidth at a central site for all remote sites. You attempt to make an initial guess at the proper oversell ratio (in this case 1:1.75 for CIR). Once the network is up, you monitor it, graphing utilization of your different sites, and adjust your oversell ratio as necessary.
Digital conversion devices
Today, most of us are familiar with modems, but how many of us really understand them? The word "modem" is a combination of the words "modulator" and "demodulator". In simple terms, a modem takes the digital data from your computer and converts it into analog tones that are compatible with the POTS. This is called modulation. The modem at the other end converts the tones back into digital data. This is demodulation. This is fine for connecting to analog telco lines like POTS, but what about connecting to digital telco lines like T1s and ISDN BRIs?
For digital telco lines, a digital to digital converter is used. This is necessary because routers do not speak the same language as telco lines. Each type of digital line requires its own digital conversion device. Each T1, E1, 56 Kbps and 64 Kbps require CSU/DSUs while ISDN BRI lines require ISDN NT-1s. T1 and E1 lines use a different type of CSU/DSU the 56 Kbps and 64 Kbps lines. When ordering equipment you must be extremely precise about what type of data line you will be attaching to it. Many analog modems will handle 56 Kbps leased lines, but not 64 Kbps. Similarly some T1 CSU/DSU only support full T1 and not fractional T1.
ISDN PRI lines require a different digital conversion device altogether. Since the router which controls the DSU functions internally, only an external CSU is needed. In this case, a CSU/DSU will not do the job, because they are an integrated unit. Their functions cannot be separated from each other. Some Cisco channelized T1/PRI cards require external CSUs, like the one for the Cisco 4x00 series routers. Others like those available for the AS5200 and Cisco 3600 series have the CSU built-in and the telco connection goes right into the router without any external digital conversion devices.
In fact, many of the new Cisco routers, such as the 2524, 1600, 2600 and 3600 series have internal digital conversion devices. These routers can be configured with an internal NT-1 and CSU/DSU (56/64/T1). In this case, no external devices are needed. This helps reduce the cost of your networking hardware, while increasing its functionality. Currently, you have to take special steps to gain remote access to CSU/DSU equipment. When they are fully integrated into IOS, you gain the ability to control and monitor them through the router. This is an extremely powerful tool for the network administrator, as only large companies could previously justify the expense of powerful CSU/DSUs with remote monitoring capabilities.
Line |
Device |
Types of External Devices Available |
56 Kbps |
CSU/DSU |
Some normal modems convert EIA-232 to 56 Kbps 2-wire or 4-wire leased lines 56 Kbps 2-wire CSU/DSU 56/64 Kbps 4-wire CSU/DSU |
64 Kbps |
CSU/DSU |
56/64 Kbps 4-wire CSU/DSU |
T1 |
CSU/DSU |
T1 CSU/DSU |
E1 |
CSU/DSU |
E1 CSU/DSU |
BRI |
NT-1 |
ISDN NT-1 |
PRI |
CSU |
T1 or E1 CSU |
Table 5-3. List of devices necessary for connecting different types of data lines.
Cabling
Cabling can be one of those things that sneak up on you and delays your network installation. Cisco has dozens of network connectors available for their routers, and those in turn attach to external devices, which may or may not have the same connector. This gremlin can be avoided with a little bit of research. For example, say you decide to order a channelized T1/PRI card for your Cisco 4500.
Your first step would be to go to Cisco's web site and look up the installation guide for the Cisco 4500 router at:
().
You would then go to the section for installing this particular card. In it, you would see the figure below, which shows the female DB-15connector on the back of the card.
Below that you would see the pin-out for the cable necessary to connect something to that card.
Just below that are the words indicating that you connect it to an external CSU. Having read the above section, you know that an external CSU is a different device from an external CSU/DSU.
The CT1/PRI interface cables have two male 15-pin DB connectors (one at each end) to connect the CT1 network processor module with the external CSU. Table 7 lists the pin-outs for the null-modem CT1 cable, while Table 8 lists the pin-outs for the straight through CT1 cable.
The Cisco documentation only mentions two cables available; both are 15-pin DB to 15-pin DB. One is null modem and the other straight through. One would assume from this that all manufacturers of external CSUs provided a 15-pin DB connector on their devices. One manufacturer is Adtran. A quick trip to their web site at will show you their specifications on the Adtran T1 CSU ACE. If you look at the product specifications, you will notice it says "Physical Interface RJ-48C: 8-Pin Modular."
You now know that a cable is needed to convert the female 15-pin DB connector on the Cisco CT1 card to the female 8-pin RJ48C connector for the Adtran. It does not explicitly say that such a cable ships with the product, which prompts a phone call to Adtran pre-sales technical support. The Adtran technician reports that a female DB-15 to female RJ-48C adapter ships with the T1 CSU ACE. This will do the conversion from DN-15 to RJ48C. You still need a male-to-male
DB-15 cable, and Cisco offers these.
Going back to the Cisco web page and looking under cabling, you will see the following.
Two standard CT1 serial cables are available for the CT1/PRI module from Cisco Systems: null-modem (part number 72-0800-xx) and straight through (part number 72-0799-xx). Null modem cables are used for back-to-back operation and testing. A straight-through cable connects your router to an external CSU (Channel Service Unit).
This clearly shows you that you want the straight through cable. Now all you need is a straight through RJ48C cable, and the connection will be complete. You can use a short RJ-45 cable, (almost any kind will do), silver satin, or CAT 5. You should really use a cable that isolates the send a receive pairs, but for very short distances, a standard RJ-45 network cable will work fine.
As you can see there is more involved in cabling than meets the eye. By making sure to investigate the equipment and spending a little time on the phone with the various vendors, you can avoid costly delays. Most, if not all the information you need is freely available on the world wide web; all you need to do is look for it.
![]() ![]() Chapter: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
Reserve your copy at a Beta Bookstore near you! |
Contact Bet@books © 1998 The McGraw-Hill Companies, Inc. All rights reserved. Any use of this Beta Book is subject to the rules stated in the Terms of Use. |