Networking Essentials Exam Guide Errata, Additional Hints and Other Comments
Being a perfectionist, I wish I could tell you that the Networking Essentials Exam
Guide went out the door in perfect condition. Unfortunately, despite the best efforts of
all involved, a few errors have been found in the 830+ pages. I will endeavor to keep this
page up-to-date with any additional information as it becomes available.
[Note that beginning July 14, 1998, I will putting a date at the beginning of new
additions so returning visitors can quickly determine what is new.]
Please send any additional errors to me at dyork@lodestar2.com.
Thank you - and good luck on the exam!
Also note that for those you are interested, I have added a web page describing the editing process of this book.
Chapters
[7/14/98] p.35 - A reader has commented
that Windows NT Server is not listed in the list of operating systems that can work in a
peer-to-peer network. His concern was question 02-03 listed on page 616 where the
question asks which operating system will NOT support a peer-to-peer network. In
truth, the list on page 35 was never intended to be complete - there are numerous other
operating systems that are not listed here (ex. OS/2, Linux, Be OS, UNIX, etc.)
If you choose to setup a machine as a "member server" when installing Windows NT
Server, that machine can participate in either a workgroup (peer-to-peer) or a domain
(server-based).
[8/28/99] p. 67 - I never noticed
before, but the table heading rows for table 3.3 are rather messed up. There also extra
words in the heading row that probably shouldn't be there. Interestingly, the data
is all in the correct columns. If I were to do it again, the headings would be:
| Layer |
TCP/IP
Networks |
Novell
NetWare |
Microsoft |
Apple |
p.69 - The "From Here..." section
would have you skip Chapter 4... probably not a good idea! :-)
p. 74 - Figures 4.2 and 4.3 don't show any nubs on the
ends of the T-connector or the barrel connector. You might wonder how the BNC cables would
lock on the connectors. Also, in Figure 4.2, despite the label, you don't get the
idea from the picture that one end is terminated while the other end is a cable. It looks
like both ends are cables - which would be a perfectly normal use of a T-connector.
A quick search on the on web brought the image below of a barrel connector. While the site
where this image was found indicates it is a 75 Ohm connector, it looks essentially the
same as the 50 Ohm Ethernet connector.
p. 95 - A really minor point - subsequent exposure to
cabling standards has shown me that manufacturers list "TIA" before
"EIA" and refer to cable as "TIA/EIA 568" versus "EIA/TIA
568." This is nothing Microsoft will test you on, but if you do look at cables,
you'll usually see that designation on a cable.
p. 97 - Figure 5.5 shows STP with a shielding around the
entire cable, with the pairs of cables running down the center. While some STP cable might
be constructed in such a fashion, you might also find it where each pair of wires
is encased in its own individual layer of shielding.
p. 99 - Figure 5.7 is once again missing nubs on the end
of the T-connector.
p. 105 - Table 5.8 lists STP as having a transmission
speed of 16-500Mbps. Not quite. Try 16-155Mbps! (as mentioned in Table 5.3 on page 98)
p.109-133 - Chapter 6 on wireless networks was included
for completeness and because Microsoft mentions wireless in the exam objectives.
However, the exam itself has been found to have extremely few questions on wireless
networks - so I wouldn't stress out too much about studying the specifics in this chapter.
Understand the concepts and terminology, yes, but you don't necessarily need to
memorize the speeds of the different wireless network types.
[11/16/98] -This isn't on the exam at
all, but some folks have now developed technologies to allow Fast Ethernet over radio.
There was a small article in a recent Internet daily email
newsletter. Not something on the exam, but interesting to think of the future
possibilities.
p. 136 - The Key Concept on this page needs some
adjustment. Network Interface cards in fact operate at both the
Data Link and Physical layers of the OSI Reference Model. The key point is
that beyond simply providing the physical connection to the network (as mentioned in the
sentence above the Key Concept), a NIC also implements the functionality of the Data Link
layer.
p. 169 - No error here, but an additional explanation
about the Key Concept at the bottom of the page. A reader, Ralph Bergquist, wanted to know more about why
RG-58/U is not acceptable for 10Base2 networks. He dug around in some communications
manuals and determined that the issue is that RG-58/U has an impedence of 53 Ohms, while
RG-58A/U and RG-58C/U have the standard impedence of 50 Ohms. If an RG-58/U cable is mixed
in with A/U or C/U cables, this slight mismatch in impedence may cause electrical signal
reflections which will degrade the quality of the network signal. (Ralph, by the way, did
pass the exam with a high score!)
Subsequent testing of the matter with troubleshooting tools has indicated that the
variation is so slight (53 versus 50 Ohms) that it is highly unlikely that any reflection
would be caused by mixing a RG-58/U cable into a RG-58A/U network. Still, as
Microsoft tests on the matter, for the exam know that you cannot use RG-58/U in a 10BASE2
network.
p. 167 - 171 - After subsequent exposure to Ethernet
standards, I have to state that I am wrong regarding the 5-4-3 rule and the maximum number
of nodes on a network (as shown in Tables 8.1 and 8.2). To be quite honest, I do not
believe this is anything Microsoft will test you on, but I want to clarify the matter for
your own understanding. In the book, I was trying to simplify much of the
information about the 5-4-3 rule and wanted to give readers a definitive number for
studying. Unfortunately, I was not correct.
For instance, the 10BASE2 rule about 30 nodes per segment is not 30 computers per
segment, but rather 30 connections per segment. That was my mistake. I
assumed 30 computers per segment times 3 populated segments equals a grand total of 90
computers per network. However, what if each of the connections on one segment were
in fact a repeater? All of a sudden you could have 30 more computers on that
segment!
The rule is that between any two distant points on your network, there can be no
more than 4 repeaters, and only 3 of the segments may be "populated" (i.e.
having more than two connections).
p. 172 - It's not something you'll be tested on, but I'm
wrong where I mention that 10BASE-T is not subject to the 5-4-3 rule. It is.
The deal is that it is still subject to the restriction that there can only be a maximum
of five segments separated by four repeaters between any two nodes on a
10BASE-T network. However, all five segments will be viewed as only
"link segments" (or "trunk segment") containing only two connections
(the NIC and the hub). In this way of thinking, none of the segments would be
"populated" - they are all just link segments.
The populated segment rule does come into play, though, when you have a network with a
mixture of 10BASE-T and 10BASE2 cabling. Then the rule is still that between any two
points there can only be four repeaters with three segments being populated.
Essentially, for pure 10BASE-T networks know that any two points on the network can be
separated by no more than four repeaters. Again, Microsoft will probably not test
you on this, but I put this information here for your own future knowledge.
p. 174 - There is a bit of confusion here in that my
bullet item at the top of the page for 10BaseFP mentions that distances can be up
to 500 meters, yet Table 8.4 specifies a maximum segment length of 2000 meters. For
the exam, I would remember that fiber-optic cable can go up to 2000 meters. The item
about 10BaseFP is in there because we found that other networking books went through all
the different 10BaseFx terms but never defined any of them. We wanted to
define them so that people would know - but Microsoft doesn't test on the different types
of fiber-optic cable - so you don't need to memorize the different fiber cable types!
p. 177,178 - The second column in Table 8.5 is headed
"10BaseT Specifics" instead of "100BaseT Specifics." Oops.
p. 178 - In Table 8.5, the formatting makes the data for
"Cable type" and "Maximum segment length" hard to read. The two lines
should have appeared as:
| Cable type: |
Category 5 UTP - 100 BaseTX
Category 3,4,5 UTP - 100BaseT4
Fiber-optic cable - 100BaseFX |
| Maximum segment length: |
100 meters (328 feet) - 100BaseTX, 100BaseT4
2000 meters (6,561 feet) - 100BaseFX |
p. 195 - Table 8.8 needs some commas to make it a bit
clearer. A quick read of the table might make you think you could go 3485 meters with UTP!
The distances for ARCnet could best be summarized in a table as follows:
| Media |
Distance |
| RG-62 A/U |
600 m (2,000 ft) |
| UTP |
121 m (400 ft) |
| Fiber-optic |
3485 m (11,500 ft) |
p. 197 - Figure 8.14 doesn't really show what LocalTalk
looks like. Not a big deal since it's not really heavily used any more (would it have
anything to do with a maximum speed of 230 Kbps???)... but I was hoping to show what the
connection devices looked like. I couldn't find any pictures and, not being a great artist
myself, had to attempt to communicate what I wanted to the QUE illustration department. I
failed. The little connection units look like a Y. They have a male cable coming out the
bottom, and two female jacks on the top. Simply plug in a cable from a Mac directly into a
connector - or plug the male cable of another connector into one of the female jacks -
presto, you've got a network! Pretty brainless and simple.
p. 213 - Figure 9.1 should have a block in the Session
Layer for "NetBIOS". Figures 9.2 and 9.3 correctly show NetBIOS functioning over
IPX/SPX and NetBEUI. It also can sit on top of TCP/IP. This is mentioned on page 222, but
not shown in the diagram.
p. 342 - The Web URL listed for the World Wide Web
Consortium is http://www.w3c.org/ . While this will
actually work, the correct URL is http://www.w3.org/ .
p. 357 - Figure 15.7 should show all the lines from the
lower Web Browser pointing to the proxy server box inside the circle for the
Corporate LAN. Someone in QUE's illustration department got a little enthusiastic...
p. 359 - The From Here... section for Chapter 15
would have you skip over Chapters 16 and 17 to go directly to Chapter 18. This is an
artifact from Chapter 15 being originally placed later in the book.
p. 415 - In the Tip, I indicate that a
continuity test with an ohm-meter across a piece of coaxial cable would register 50 ohms.
Incorrect. In fact, it would register 0 ohms. You should get 50 ohms if you were to put a
terminator on one end, and then touch one lead on the ohmmeter to the center core and the
other to the outside shield. That circuit would register an resistance of 50 ohms due to
the terminator. I meant to indicate that if you used a cable test specifically designed to
measure impedance, you would get a 50-ohm reading from the cable. However, when you
typically refer to an "ohmmeter" it is for continuity testing.
[7/14/98] p. 445 - One Gbps is
essentially one billion bits-per-second, not one million as stated in
the text! Not quite sure how that snuck in... perhaps because my eyes became so
glazed over when we were in the final proofreading stages!
Test Questions
p. 577 - Self-Assessment Test - Question #8 has a
scenario with the proposed solution involving Thinnet. The required result is that the
network should be as inexpensive as possible. Optional results are that the network should
be easily reconfigurable and should be easy to install. The book correctly identifies the
answer as B, the solution meets the required result and one of the optional results (easy
to install).
Several readers questioned whether 10BaseT would actually be cheaper. The answer is that
it would truly depend on the number of computers to be networked. With Thinnet, you can
typically pick up T-connectors for $3-5, cables for around $10, and 2 terminators for
around $5-7 apiece. Many NICs for Thinnet include the T-connector along with the NIC. So
if you had, say, 4 computers, your network media (forgetting about the cost of NICs) might
cost you somewhere around $30-50.
On the other hand, the same network with 10BaseT would involve purchasing a hub at around
$60, plus shelling out $5 per 10BaseT cable for a total cost of somewhere around $80.
This assumes you are running your cables in a relatively confined area and using 8-10 foot
cables. If you start to run longer distances, 10BaseT has, in my experience, winds up
being cheaper per foot. If you're interested in exploring pricing issues, there are
several sites on the web. One that I know of is the DataComm warehouse at http://www.warehouse.com/DataComm. (Note: I
do not intend this as an endorsement. I know nothing about this company other than the
fact that they shower me with junk snail-mail catalogs hawking their wares. If you know of
other sites of value, I'll be glad to add them.)
So, to be more precise, I should have started the question "You are in charge of
the four computer systems used by a small nonprofit organization...." and
provided some quantity for the question.
[7/14/98] p. 616 - Chapter 2 - Question
2-03 is correct. Some readers have challenged whether Windows NT Server should also be
correct. As noted in my response to p. 35 at the top of this page, Windows NT Server
can be used in a peer-to-peer network provided it was installed onto a machine as
a "member server".
[8/28/99] p. 630 - Chapter 4 - Question
4-04 has the correct answer of A (repeater), but one could argue that answer B (hub) is
also correct because... gee... what's a hub? It's a multi-port repeater! I'm
surprised no one ever pointed this out before. Still, a repeater is the best
answer.
p. 630 - Chapter 4 - Question 4-06 states: "A
network based on a bus topology has many advantages, including which of the following?"
The answers are correct as printed in the book: B. Easy installation, C. Cabling is
inexpensive, and D. Easy to expand. However, choice D is perhaps a bit vague.
If you are at either end of a bus network, it is very easy to expand the network. Simply
remove the terminator, add another length of cable and a T-connector, and terminate the
new end of the cable. Simple. Easy. Of course the network will be down during the brief
time period when you have the terminator off the cable, but other than that, it is easy to
expand the bus network. On the other hand, if you want to reconfigure the network to add
10 computers into the middle of the network, it can be a bit of a pain to find where to
add the cable and to make sure that everything is connected properly. In contrast, it is
extremely simple to expand or reconfigure a star network, as it merely involves
adding additional cables to a hub.
[8/28/99] p.633 - Chapter 4 - Question
4-12 asks which of the topologies would minimize network disruptions. The answer (A. Star)
is correct, but answer D (Mesh) could also be correct, as a mesh network
typically has redundant connections that would avoid disruptions. However, as mesh
networks are usually extremely expensive, they are seldom implemented on a LAN level.
Just for clarity, there should probably be an extra line in the question that says
"You also want to minimize the cost of the network."
p. 638 - Chapter 5 - Question 5-11 states: "You
need to add additional computers on to your existing Thinnet network. You send a coworker
out to purchase additional network cables. He returns with RG-59 cable. Will this work?"
The answer stated on page 721 and also on the CD-ROM is "B. Yes, but performance will
suffer."
While I believed this to be true in some testing I did, I have since been advised
by some who work with cable that I am incorrect on this. The correct answer should be
"C. No, it will not work."
And then again... in the development of a course I am doing now on
troubleshooting Ethernet networks, I had the opportunity to spend some time with some of
the top engineers at Fluke, one of the manufacturers
of network and cable test equipment. The engineers demonstrated that yes, indeed, we were
correct in the book. RG-59 will work, but due to the reflections created by
the change in impedence performance will definitely suffer.
If we were to do the book again, I would just leave this question out, because, while it
is a "real-world" example, it's more of a trick question and is not likely to
appear on the Microsoft exam in any event.
p. 646 - Chapter 6 - Question 6-10 asks: "Which
of the following are frequencies allocated by the FCC for unregulated use? (Choose all
that apply.)" The answers on page 721 correctly indicate choices B and C, but
several readers have written stating that they believe that choice D (5.72-5.85 MHz) is
also correct. If you look closely at the bulleted list on page 114, you will see that the
third range allocated by the FCC is 5.72-5.85 GHz, not MHz. Yes, it was a
bit tricky of me... but that is how Microsoft will be on the exam!
p. 653 - Chapter 7 - Question 7-08 asks which IRQs can be
used for a NIC given a certain scenario. The answer on page 722 (B. IRQ 9) is
correct, but choice D (IRQ 15) could also be correct in certain circumstances. Choice A
(IRQ 3) is out because of COM2 and C (IRQ 10) is out because SCSI controllers typically
use IRQ 10. However, the scenario merely states that two hard drives are used, but not
whether two hard drive controllers are in use. If both drives are on their own
IDE hard disk controller (HDC), the answer is correct as it stands. The primary HDC will
be using IRQ 14 and the secondary HDC will use IRQ 15. However, in most computers today,
you can attach two IDE hard drives to a single ribbon cable connected into the primary
HDC. In this case, IRQ 15 would be available because the secondary HDC would not
be in use.
The question should have been more precisely worded: "It has a serial mouse, a
modem on COM2, two hard drives (each on their own controller), and a SCSI controller for
additional SCSI drives."
[9/7/99] p. 661 - Chapter 8 - Question
8-09 just shows how times change. Two years after the book was written no graphics shop
would want to be at 10 Mbps - they all want to be at 100 Mbps! No error in the
question, just a comment.
[9/7/99] p. 663 - Chapter 8 - Question
8-14 begins "AppleTalk uses..." It should really begin "LocalTalk
uses..." as the intended answer is for CSMA/CA. AppleTalk protocols can
now work with Ethernet (EtherTalk) or Token Ring (TokenTalk).
[9/7/99] p. 665 - Chapter 8 - Question
8-19 is just plain wrong. The answer is supposed to be D, I assume because with 170
employees it would exceed the 90 computers I once mistakenly though was the limit on
10BASE2. However, the correct answer would be A, the solution meets the required
result and both optionals.
Here's how the network could be layed out. Let's assume the old network was two
segments of 25 separated by a repeater. For the new network, we buy a second repeater and
connect both segments to a new "backbone" segment. Presto... our old network is
still operating fine. Now we extend the backbone segment to the floor above us. Up
there, we buy five repeaters and attach five segments to the backbone separately using the
repeaters. Why five segments when we have 120 people? (i.e. shouldn't 120 /30 = 4
segments?) Well, there's the matter that one of our thirty connections on a segment is to
the repeater. So each segment connected to the backbone will have 1 connection for the
repeater plus up to 29 client computers.
Now does this violate the 5-4-3 rule? Take a look at it... the backbone segment has a
total of 7 connections (2 downstairs and 5 upstairs). If you take any two computers,
you'll find there are only 3 segments and 2 repeaters between the two points. (the local
segment, repeater #1, the backbone segment, repeater #2, the remote segment)
So in the end, the solution would work perfectly fine!
Last Modified: October 01, 2008
|
