Well, I took the TCP/IP test yesterday and did fairly well. Got an 857.

Thanks for all your suggestions and help. I had heard that this test has a reputation as being the hardest test out there but found it to be relatively fair. Here's my take on it for all you future TCP/IP test takers:

Study up on SNMP. There are several questions about it.

- Be able to jot down the subnetting table.

- Know how to create and modify routing tables for NT routers and how to modify them when they're on different *custom* subnets (for example, if your subnet mask is 255.255.192.0, know how to define the mask in the routing table if you're pointing to a subnet with a mask of 255.255.240.0).

- Know what order TCP/IP will search for NetBIOS name resolution. This was actually a VERY easy question. I wasn't sure of what order it was but the question was phrased so that only one answer could possibly be it. It was something to the effect of:

1. check the LMHOSTS

2. check the cache

3. IP address of host is resolved and delivered to source.

4. source sends broadcast.

Okay, that was a *real* rough paraphrase but as you can see, option 3 has to be the last step and only one of the multiple choices had that as the last step. Voila! *Many* questions were like that....where *one* little thing would clue you in on the answer.

- Know more about Performance Monitor than anyone really should. (Such as the difference between Object:IP Datagrams/Sec and Object: TCP Segments/Sec.).

-Blaine

Guys, and Ladies,


Just like everyone know that i have just passed the exam... after lots of hard work and study hours...

Total # of questions: 58
# of questions needed to pass: 45
passing grade: 750
Time allow: 90 mins

lots of utilities questions: make sure you know the following:

network monitor, performance monitor, netstat, nbtstat, arp, route, ping, tracert, ipconfig... etc.

lots of questions on connectivity: for example, why does one computer in subnet 1 can ping a host on subnet 2 with ip address, but not with a domain name? (and variation of it)

senario questions includes: you have subnet 1, 2, and 3, why can hosts in subnet 1 & 2 can talk to each other but not with 3, you are using wins, and router is working correctly. (variations of it)

rating questions: make sure you know, snmp, dhcp, wins, dns, because at least 4 questions asks you whether certain configuartion will work or not, and you have to reply either: it works and meet all requirements to it does not work...

subnet masks calcualation: make sure you write down the table before you actually start writing the exam as it will save you time... these questions are straight forward, nothing fancy. about 4 questions was based on it.

wins partner: pull and push, one question ask you that you have 2 subnets, 1 in dallas, 1 in seattle, and it ask you to make sure all hosts in both subnets can talk to each other, and it gave you a chart stating the config of the wins server in both subnets... you have to rate whether the config will work or not... i answer this one e) dont work.

know hosts and lmhosts, differences and similarity.

know dhcp relay agent, dhcp manager, and dhcp options, client reservations etc.

make sure you know that if you have 1 hosts in 1 subnets and the other host in another subnets and both are connected by router, those might have incorrect configured subnet masks which leads to the problem that they might not be able to talk to each other even if they correctly specifed an ip address for the router.

know something about the PPTP, one questions was ask about it.

know that performance monitor can record data and allow you to save the data into a file for later analysis or import to a spreadsheet or database program.

know the fact that netstat start recording network staticstic right after you got the server (computer) running, the statistic is cumulative.

know how to configure routing table, multihome, static & dynamic routing, (very important)

nothing on igmp, sligting windows, nslookup...

if you study the area which i mention above, you should have a very high chance in passing the exam. overall, i found the exam tough and hard, but if you know your stuff, i guess is easy. for me, since i receive no professional training on it, i spent a whole month preparing for it.

please feel free to add on to my comment or correct me if needed.
and if you like to thank me for writing this note, you are welcome, some encouragement is always needed...

Alex


Just passed 70-59 Internetworking with TCP/IP on NT4. Scored 948, so I'm now 2/6.

I found the exam very tiring so my advice is to arrive fresh - there's 59 questions in 90 minutes but there's a lot of reading to do - some long questions.

I got some great info from W. Brian Talbert's pages at http://members.aol.com/wbriant/index.html

Some observations on 70-59 ...

The subnetting questions were all straightforward. I used the advice from the list to write down all the mask information I could before starting the exam.

You've gotta know the name resolution processes for DNS and NETBIOS.

Many questions on troubleshooting these. Learn the format of the LMHOSTS file. Know how WINS clients communicate with WINS servers and understand the WINS database replication process.

Be familiar with DHCP scope options and which can be overridden by the client. Understand how to make NT route packets, how to make NT dynamically update routes, and how to provide DHCP resiliency across routers.

Brush up on TCP/IP printing using lpr and be familiar with SNMP installation on NT. Know which troubleshooting tools are used for which purpose.

I'd like some advice please as to which exam I should take next, I've done NE and TCP/IP and my background is mainly networking with some NT administration. Would it be a more effective use of study time to take 70-68 (Enterprise) ahead of 70-67 (Server), or vice versa? Would it be better to take Exchange after these 2?

Thanks & keep that advice coming!

 

Andrew


Took tcp/ip for 4.0 5/12/97 and passed with 775 of 1000, pass mark of 750 and 58 questions. Prepared with 3.51 material and New Riders study guide.

From previous post to this list I expected app. 40 % subnetting and IP type questions however I only had about 12. The tip to memorize the custom subnets and number of host per saved time , and I would suggest practicing creating your table from scratch a few times pre-test. The often mentioned tip on scenarios with excellent, acceptable, poor and will not work choices ( majority of which were will not work ) did not show up at all.

The empahasis on the exam I wrote was name resolution with several difficult questions of the type Host A can ping with IP but not Name - Why ? This was iterated continuosly with differant DNS , WINS, NetBios, IP , remote vs local network combinations.

Also got about 5 questiond on IIS as well as primary vs. secondary vs. caching-only DNS servers. Only 1 push-pull scenario.

All in all , I made this test tougher then need be by prepping from 3.51 material then changing to the 4.0 exam at the last minute, but alls well that ends well.

Anyway thats 5 down and IIS scheduled for 5/19. Thanks to list members for input. See Ya !!!


Hello All!

I recently (last Friday) wrote and passed the TCP/IP exam. This was my fourth exam and definitely one of the hardest I've taken. I scored 780.

I don't have an extensive background in computers or networking and I'm a math-phobic, so I was happy just to pass, although I wish I had done better. I think I probably got tripped up by some of those famous Microsoft 'trick questions.'

For preparation, I used the New Riders book and the Transcender (both were very helpful although the book could use some editing). I also used a lot of material out of TechNet. For subnetting, I did a lot of practice questions (these and the subnetting table allowed me to score well on that section of the exam). The outline of the exam provided on the Roadmap CD is an accurate list of topics to expect.

Many people on the list have mentioned being pressured for time on this one. That wasn't a problem for me. I went through the exam in about 30 minutes and had plenty of time to work on the more challenging questions and go over all of my answers again thoroughly.

My strategy on the exam included memorizing a subnetting table, the IP address classes and some other general information and doing a brain dump on paper immediately before I wrote the exam. I never once used the calculator and I found that the subnetting table was adequate for all the subnetting questions that I encountered (and I apparently got every one right). For the multiple rating questions, I answered each one 'e' (doesn't meet the requirments and doesn't work) and then marked it. I went back through these after I had gone through the test one time and changed only two of them to 'a.'

One of my greatest concerns prior to the test was the point-and-click and write-in-the-answer questions. From what I've already read on the list, I'm beginning to suspect that everybody is getting the exact same ones on every test. I received three. One showed a diagram of a network and asked me to click on the computer that should be the Wins proxy. Another showed me the results of an ipconfig /all command and asked me to locate the problematic configuration. My write-in-the-answer question presented a troubleshooting situation and asked for the command to identify the problem. None of these were really difficult,but I didn't feel like one of the point-and-click questions allowed you to pinpoint the answer accurately enough.

There were a few freebie questions on this exam, but for the most part, each question did more than ask you to regurgitate information. Many questions integrated more than one topic and expected you to have a solid understanding of all of the concepts. This was especially true of the troubleshooting questions. You can memorize trivia until it comes out your ears, but it's of no use unless you're able to put it all together in a coherent way.

My biggest downfall? I had a few horrible, convoluted questions involving printing in heterogenous environments. The one that really gave me problems involved a unix machine, an NT machine, TCP/IP and DLC and who can print to who doing what...ARGGGH! A few questions had more than one plausible answer and there were several that looked like obvious trick questions. Before tackling this test, make sure you spend some extra time on sections of the books that cover heterogenous environments (ie: unix, netware). There was a big emphasis on this on my exam.

All in all though, this exam wasn't nearly as bad as I thought (or rumors on the list had led me to believe)! I expected horrible exhibits and most of them were actually pretty similar to what was on the assessment. I thought that the subnetting questions were the easiest part of the exam. I had five questions on SNMP and they were very straightforward. Know how telnet and ftp work. Make sure you understand NetBIOS and Host name resolution thoroughly (I memorized these and wrote them down before the exam).

Here are my stats for the exam:

Installation and Configuration 75%
IP Addressing 60%
Subnet Addressing 100% (Yay!)
Implementing IP Routing 75%
IP Address Resolution 100%
Host Name Resolution 50%
NetBIOS Name Resolution 75%
Connecting to a Heterogeneous Env. 60%
Implementing the SNMP service 100%
Performance Tuning and Optimization 100%
Troubleshooting 83%

Thanks to all of you, I'm moving right along! Four down, two to go! If anybody has specific questions about the exam please email me - I'd be glad to help.

Thanks for your help and advice, everybody!!


Jayme

Hi, all, I've just passed the TCP/IP exam, 70-53 this morning.

Thanks to all the tips and advise I got from this list, thanks again, without you people, properly not possible.

I used New Rider's 'TCP/IP and SMS' book and Telnet and Transcender.

Anyway, I've passed with 857 and 715 is required, here's some questions that I can remember and worth pointing out, some of you may have got it already, well, another person, another test ... Ha.

1. Like many people said, select 'E' (does not meet the requirements and does not work), for those MRI and marked them and reviewed them later on, I've only had 1 question that's a or b.

2. The fill-in-the-blank question for me is to answer 'ping market'.

3. Clearly know the different between NetBIOS name and HOST name, very important.

4. HOSTS. is case-sentsitive, LMHOSTS. is not, remember!

5. #PRE is only for LMHOSTS, not HOSTS.

6. The point-and-click question for me is to click on the [enable routing] box in the Advanced TCP/IP setting.

7. Lots of exbhits questions, be perpare, it's very difficult to look at some of the ip address in the pictures.

8. Ping a host, a host has to reply, remember, so, the remote host's gateway has to be configured. I had a few questions that the remote hosts haven't got a gateway, that's the problem, but of course, only if the remote host hasn't got any satistic routing.

That's all I can remember at this time, I may publish more tips on my home page, check it later: http://shell.ihug.co.nz/~bulldogs/

Thanks.


I passed this exam back in the fall of 96 (846, 714 required).

The most important thing in this exam is knowing the math- be able to compute *rapidly* whether two machines are on the same network, etc. I would strongly advise taking the little practice exam, and ignoring it in order to take a moment to write out on the scrap paper some of the subnetting math in chart form (you've got a free 10 minutes-take it).

I only studied a week for this (and only about 30 minutes a night), so was not as quick as I could have been. Time management is crucial, even if you know the material-save all questions that require comutations for the end. Don't waste a single minute second guessing- there isn't enough time (this sentiment was echoed by many experienced people I've spoken with, as well).

You want to make sure you get to all the ones you KNOW the answer on. If I had another 15 minutes, I would have another 40-50 points, as there were questions I could easily answer, but didn't have the time to chug my way throught he calculations (should have had the math a bit more "automatic"). Nothing tricky, though. Interestingly enough, all but one of the "problem based" questions had "e" as the answer; didn't fill the reqirements, didn't work.

Good luck.


Passed the exam first go yesterday, after attending the course last week. I thought the exam was very fair but contained several very ambiguous questions, especially a couple that required more than one correct answers to do with DNS and host name resolution.

I think of the 58 questions, it was splitted up as follow:

30% tcp/ip utilities, eg.. ftp, netstate, arp, network monitor etc...
30% DHCP & WINS
20% DNS and name resolution
10% TPC/IP printing, SNMP
10% Subnetting, know your tables well

Concentrate on Subnetting, DHCP, WINS, DNS and you will have virtually passed the exam.

Good luck all

JJ


I just got back from the new tcp/ip test and I wanted to pass on some information.

58 Questions, 90 minutes, 750/1000 to pass

  1. Know all utils and what they are used for (arp.exe, netstat, nbtstat)
  2. Know when to use performance monitor, network monitor, netstat to perform various functions (ie when would you use one over the other) (4 questions on this)
  3. Know the node types: h-node, b-node, when to use which
  4. Know all the types of DNS server, primary, secondary, cache-only
  5. TCP/IP printing I had about 5 questions on this subject
  6. Troubleshooting various IP situations. Most of these involved either wrong gateways, no DNS server, wrong node type, no wins server.
  7. Know when you need a DNS server vs when you need a WINS server
  8. Subnet questions - I had around 6 to 8 questions on choosing a subnet for the proper number of hosts. If you know the table on this months MCP Mag web site you will be just fine.
  9. I had only one question dealing with route add
  10. Know what a CNAME is to a DNS server
  11. Know about WINS replications, when to use push vs pull
  12. Know how to properly configure a lmhosts file

Overall a fair test

Thanks, and good luck!


1. Browsing, a bdc is a backup browser, if the maintainserverlist is yes, the computer is a backup browser, else the master browser will select a computer, if required, to act as a backup browser with a basis of one browser per 31 per computer plus one in the domain/workgroup. Two questions.

a) What criteria are used to pick a computer, is it just the next computer to come online that can act as a browser?

b) Is it the number of currently active computers that determine the number of backup browsers required or the number registered as belonging to the domain.?

 

2. Name resolution and Node types

What determines the node type. it it the options ticked in the tcpip control panel for NT clients.

Maybe someone could help me fill in this matrix

  Nothing ticked enable lmhosts only enable DNS only DNS and LMhosts
 wins yes  p (hybrid?)  hybrid  no  broadcast

If my client is wins enabled, is it automatically hybrid (h)?
if not and enable lmhosts /dns not ticked is it automatically broadcast (b)
When is a node an m type?

Well, I finally did it. It's been three months since I began my quest for the holy grail of certification; I finished up my MCSE this afternoon by passing TCP/IP for NT 4.

  1. You can NET USE a server, but can't connect to its IP address. What's the problem?
  2. You can ftp to the ftp server's hostname, but not to its IP address. What's the problem?
  3. You have a net in Seattle & one in Dallas, how should the WINS replication be set up?
  4. You need dynamic updating of computer name to IP address mappings, what do you use?
  5. Effects of your host having an invalid subnet mask.
  6. Effects of your host having an invalid default gateway.
  7. Effects of the destination host having an invalid subnet mask.
  8. Effects of the destination host having an invalid default gateway.
  9. If you have the Class C subnet table memorized, and can figure the Class B table quickly, you get six to ten freebies.
  10. Be able to troubleshoot an ipconfig /all output. The actual question on this that I had was very obvious.
  11. I can't give the exact questions (mainly because I can't remember them), but know how, why, and when to use NETSTAT, NBTSTAT, PING, ARP, etc. I didn't have any questions about command line parameters, but YMMV.
  12. Know the capabilities of PerfMon, NetMon, Netstat, and Nbtstat (as far as which does which jobs). If you want to print a chart of stats, which do you use?

That's all I can remember for now. Thanks to Herb and everyone else for your help. I will, of course, be staying on the mailing list for awhile, as I consider pursuing the MCT. Hopefully I will have the time to be a bit more active than I have of late. (If this mail goes through, that is....)

Thanks again to all. I have to go celebrate now.....


  1. I don't remember any questions on CIDR.
  2. There were a number of questions about network monitor. You need to know when to use Network Monitor and when to use Performance Monitor.
  3. I don't remember any questions about IPv6 either.
  4. There was one question where you had to supernet - create a subnet mask for multiple networks combined, but CIDR wasn't mentioned.
  5. There were plenty of questions about DHCP and WINS.
  6. Know what DNS is for and what WINS is for and how to make them work together.
  7. There were a few questions on RIP.
  8. You have to know how to create static routes for a simple network.
  9. There are questions that ask you when to use ARP, NETSTAT, NBTSTAT and PING.
  10. Remember you need to have SNMP services installed to be able to use Performance Monitor against TCP/IP.
  11. There were a number of questions that asked about situations where DCHP was needed on multiple subnets, some where there were multiple DHCP servers working together supporting more subnets.
  12. I think there was one question where the scenario said that a server could communicate with machines on another subnet, but that server had an address that wasn't correct for its subnet, at least that is what I thought. I scratched my head on that one for too long before I picked an answer and later wrote comments about it.


Awhile ago, someone suggested that you sit down and write this table on the work sheets before you started taking the test.

I agree - I wrote

nets stations

128 128

64 192 2 62

32 224 6 30

16 240 14 14

8 248 30 6

4 252 62 2

2

(sorry, I cannot reformat this to something you could read... Horshack)

 

I found there were questions that also needed class B masks

   nets stations
255.0 254 254
254.0 126 510
252.0 62 1022
248.0 30 2046
240.0 14 4094
 224.0 6 8190

With those tables on my sheet, there were a number of questions I could answer quickly.


This is the first beta exam I have taken where I worried about the time on the clock. I took the NT 4.0 Server and NT 4.0 workstation betas and passed with plenty of time to spare.

explanations of their shortcuts. They could use a lesson in subnetting
from Herb. There also seem to be 2^n (read that as 2 to the n) places
where they used the value 2^n instead of 2^n-2 in tables indicating the
number of addresses or subnets in a range.

[I missed this post when I first received it, so I am late replying]

Thanks for the compliment. Would you please confirm what I understand you to say the MOC indicates:

The number of SUBNETs are 2^n (where n is the number of 'additional' bits beyond the default mask).
Example: Class B, 255.255.128.0 mask would provide 2 subnets (this would have been meaningless under the old rules that Microsoft used on the 3.51 exam.)

Surely they don't disregard the '-2' for the number of hosts. Example: Class C, 255.255.255.240 mask would provide (2^4 - 2) hosts, or 14, RIGHT?

Allowing the use of all subnets is consistent with Microsoft's support for RIP version 2 which supports 'extended network prefix' addressing.

Allowing the use of all host numbers would STILL interfere with the multicast address on the subnet.

Douglas, you said that you might provide an update AFTER you completed the exam. If you did, I missed it. Could you please post it.

My update to your expectations includes the following:

Only 1 CIDR questions (on Beta) which means most people will NOT get a CIDR question. NO IPv6 whatsoever LOTS of DNS and DNS/WINS integration. Harder utility and LPR/LPD questions

People have been asking which is easier, the infamous 3.51 or the 4.0. I think that starting from SCRATCH the 4.0 would be easier but consider a couple of points.

1) If you understand subnetting that takes ALL of the sting out of the 3.51 test

2) There is much more information available on the 3.51 exam

I think they are about equal for me. But I understood and passed both comfortably.

I'll be taking the exam on Thursday. I don't have any info on what will be on the exam except what is available from Microsoft on the web. That information doesn't make the test seem as intimidating as it might be. There doesn't seem to be anything on integreation with the Normandy product line. I expect quite a few questions on DNS, DHCP and WINS. Probably a couple of questions on CIDR and IPv6. I also have a feeling that there may be a question about Network Monitor.

As review, I intend to set up DHCP, WINS, DNS, RIP, IIS, TCP/IP services, and whatever I can think of up a couple of times on my test systems.

It may not sound fair (it's one of the benifits of being an MCT) but I'm using the instructor's kit for the new 5 day TCP/IP on NT 4.0 class as a study aid. It's actually pretty good as far as technical course materials go. But there are a few places where I've had to scratch my head over explanations of their shortcuts.

They could use a lesson in subnetting from Herb. There also seem to be 2^n (read that as 2 to the n) places where they used the value 2^n instead of 2^n-2 in tables indicating the number of addresses or subnets in a range.


Thought I'd share what I learned from taking the exam.

First off, I didn't have any questions with more than one Exhibit. No more shuffling screen shots.

They got rid of those terrible "rate the solution" questions.

Those have been replaced with the type that other tests use where you have a required result(s) and optional results. You have to answer how many are satisfied from the given solution.

There also seemed to be less subnetting questions. The ones I had were easy. The only one that caught me off guard was one where it gave you 3 Class C addresses and asked for the mask to make them one big network.

The DHCP questions were straight forward, only new thing there is the

DHCP Relay Agent which is pretty simple.

I didn't know DNS very well so that's where i'm sure I got the lowest score. They don't ask a lot on it, but you should know it. They asked what kind of DNS server to setup to resolve Internet names, and how to setup backup DNS servers.

More host name resolution questions than before. They ask a good many on what happens if I can map a drive in Explorer, but can't get there other ways...etc...

The usual 5 or so questions on TCP/IP printing. Nothing technical, just basically what services to install.

No technical questions on DHCP or WINS, such as how to backup/restore databases, etc... Only 1 or 2 questions on setting up WINS replication.


Total Questions: 141

All in all I'd say this test is less technical than the last one. They ask more general questions now, less which box to check for this or that.

I'd be happy to answer any other questions you have about the test.

I think that this one was better than the previous IP exam (or at least has the potential to be)

There was only one question I can remember on CIDR/Supernetting. Not one question on IPv6. I was expecting that it would at least show up as an option in an answer somewhere.

Definitely a lot of questions on DNS but if you know the record types, server types and some trivia it's not so bad.

A lot of questions on what tool (tracert, ping, arp, netstat, nbtstat, perfmon, netmon etc.) would you use to diagnose a problem or display information.

A number of questions on name resolution.

One of the things that I liked was that they used the same 4 network diagram exibits over and over. That means that I didn't have to study each diagram in detail. It made a number of the questions go much faster.

There were a number of poorly worded questions. I expect that these will be cut from the pool before the exam goes live. 10 minutes of comment time is definitely not enough.

There were also a couple of questions that asked about what subnet mask you would use if you had for example exactly 4 or 8 subnets. The answer to these questions depends on if you are willing to use the all 0s or all 1s subnets. A lot of people (the test writers I suspect included) are taught that these are invalid. They aren't invalid and will work if all of the devices on the 0s or 1s subnets (most devices made today will) support those subnets. Because of that, I consider these poor questions. Depending on your point of view, two different answers are correct.

Something I would have liked to have seen on the exam would have been a question on non-homogeneous subnetting. For instance a question that would require you to break up a class B into 14 12-bit and 30 8-bit subnets. Something like that would have really seperated the men from the ... strike that, don't want to be sexist, how about MCSEs from the CNEs. Questions, questions, questions. :)

Thats the only CIDR type question on the test.

NO IPv6. Why ask something NT doesnt support?

Few on IP Router. Basically how do you make NT have dynamic routing.

Name resolution was like "How do you make Unix clients use WINS for names not in the DNS"... "How do you make Windows clients use DNS for names not in WINS"... "I can ftp to this NT server, but I can't connect to a drive from Explorer"...that type.

No serious DNS record questions. I had one on CNAME and one on MX records. The rest was "types" of servers, secondary, primary, backup...etc... Wish I had looked at that before I took the test. :)

Jason


Subject: MCSE: Just got back from TCP/IP for NT4 Beta exam.
Thought I'd share what I learned from taking the exam.
First off, I didn't have any questions with more than one Exhibit. No more shuffling screen shots.
They got rid of those terrible "rate the solution" questions.

Those have been replaced with the type that other tests use where you have a required result(s) and optional results. You have to answer how many are satisfied from the given solution.

There also seemed to be less subnetting questions. The ones I had were easy. The only one that caught me off guard was one where it gave you 3 Class C addresses and asked for the mask to make them one big network.

The DHCP questions were straight forward, only new thing there is the DHCP Relay Agent which is pretty simple.

I didn't know DNS very well so that's where i'm sure I got the lowest score. They don't ask a lot on it, but you should know it. They asked what kind of DNS server to setup to resolve Internet names, and how to setup backup DNS servers.

More host name resolution questions than before. They ask a good many on what happens if I can map a drive in Explorer, but can't get there other ways...etc...

The usual 5 or so questions on TCP/IP printing. Nothing technical, just basically what services to install.

No technical questions on DHCP or WINS, such as how to backup/restore databases, etc... Only 1 or 2 questions on setting up WINS replication.

Total Questions: 141

All in all I'd say this test is less technical than the last one. They ask more general questions now, less which box to check for this or that.

I'd be happy to answer any other questions you have about the test.


Finished the TCP/IP Exam this AM with an 815. 5 down, one to go!! Thanks for all the help from the people on this list!!!

Next up is IIS and any suggestions on that would be greatly appreciated, starting with whether the MS Training guide is worth purchasing and whether that will prepare you completely. If that does not work, are there others that are better? I checked out the table of contents, and it did not list Index Server, which I heard was an important part. Does anyone know if the guide covers that, or if you need other materials?

Now for a debrief.

I did all preparation self study, using the New Riders book to learn it and Technet to supplement and add where they were unclear or where I felt I needed additional information, or to have the material presented in a different way

I have heard people say that this one of the toughest exams and others say it is not too bad. My opinion is both are right !! The exam is tough and handling all the exhibits is a bear. I had taken the assessment exams on a 10 inch laptop screen, hoping to get used to a smaller screen, but nothing could really prepare me for the real thing, with exhibits that call others exhibits and tiny little IP addresses. Your mind is definitely frazzled after that!

On the good side, the exam content is pretty straightforward. Compared with the NT exams where you have to know something about a LOT of topics, this one you really just have to concentrate on a few . But be warned, you really need to know those topics inside and out. One of the things that I did was to copy the MS Preparation guide and then list everything I knew about each of the topics. Then, study from that.

I wrote out the New Riders chart for subnetting, but then only used it once or twice. I guess I expected more "you want to design a Class X network with Y Networks and Z Hosts, what should you use/" questions.

I echo what has been said before in terms of what concentration areas, but would add these comments:

MRIs: Lots of them and most don't work -- those that do work (and are not totally off the wall) are tough to Rate

Many different flavors of routing, i.e. which host can communicate with what, based on different routers and configurations.

SNMP -- 4-5 questions, but VERY straightforward. I thought they were the easiest part of the exam

Understand the Dialog boxes and screen and what they do for DHCP Manager, Installation, SNMP -- more of these type questions than on others exams

Did not have to know the detail switches for things like Arp and NBTstat.

Did have a ROUTE ADD question that was pretty easy.

Point and Click: I got three (locate a WINS Proxy, Bad part of an IPCONFIG command output and Printer installation)

Printing: Much tougher than I thought it would be. You need to understand a lot about how UNIX Hosts print. (Does anyone have any inkling why MS thinks printing is so important? It was also stressed on the WIn 95 and NT server exams)

Hope this helps. By the way, I may take the TCP/IP beta exam later this week. I can't imagine knowing the intricrate details of TCP/IP more than I do right now.

Chris


Passed the exam today with 870. Thanks to all those who posted info on the mailing group.

As has been previously said, the exam is not difficult itself but understanding the questions is ! The exhibit graphics are appalling quality and some questions have two of them!

As to the test itself, I started the test with the ploy previously stated of answering E 'does not work' to this type of question, of which there were about 12 and then marking them.

I then finished the exam with about 45 mins to spare and had time to go back and check all of my answers. Only 1 of MRI questions was not E.

Other than that plenty of questions on routing tables with and without RIP routers.

One question on performance counters - source quench

A couple on SNMP

Two on WINS with one on replication.

A couple on DHCP server configuration

About ten on IP and subnet addressing.

About four on LMHOSTS and HOSTS files (mainly to do with the syntax).

About five on NetBIOS name resolution (b-node, p-node etc)

A few on ARP'ing and ARP tables (including situations with machines with multiple NIC's)

The trickiest questions were about printing for NT with UNIX hosts, with and without DLC installed, sharing printers, starting the lpd service etc.

That's all I can remember, SMS next (where's the tips ?).


Well Dynomiiittte!!!

I passed the TCP/IP test today. It is still a bit of a blur but in keeping with tradition here is my rambling debrief on the subject. I won't repeat what has already been said here on the list, so don't expect a big debrief.

I had a lot of Multiple rating questions i.e.. How does this solution address the problem? (Approx. 15)

I had one where you had to place a cursor over a check box (Something to do with routing)

I had one that I had to type in the answer

The rest of the test had the normal compliment of select best answer or all that apply.

And you have all heard it before but this one can be a bit of a time management issue. Some of the questions are quite long and require a lot of comparing values. I managed to complete the test with about 10 minutes to spare, enough time to review my answers but that had my pea-brain whirring.

Custom subnetting wasn't as big an issue on my test, I thought there would have been more, I guess I was expecting the worst. Still the advice of memorizing the table and writing it down on a piece of paper before you start the test is a good idea.

Also try to test at a center that has a 17" monitor. Some of the IP addresses, etc. on the graphics would have been brutal on a 14" monitor.

 

Tips;

 

How was it?

Well I overstudied. It was easier than I thought it might be. But it is still an "Aggressive" test. I think that there was less material to memorize than Win 95, but more thought had to go into answering some of the questions. It is certainly one of the hardest I have done, but within reach if you know how to put it all together.

 

Thanks to;

Herb Martin, how do you do it? Marc Houde, who sent me some helpful material. And nameless others who made it seem the test was achievable.

I think my e-mail troubles may be solved, so here's another attempt (more than 4 failed since the first time!) to post the following:

(PS. Did not use Windows Calculator during the test at all.)


Just passed TCP/IP 3.5/3.51 (exam 70-53) today with 857. Thanks to all who post on this list. (This is my first post!) I've put my comments and reactions on the MCP Magazine forum relating to this exam. See http://www.mcpmag.com/scripts/dbml.dll?template=cfo/confer/threads.dbm&CID=2 and select the item "Exam Discussion (Exam 70-53)" under the exam's title "Internetworking MS TCP/IP on MS Windows NT 3.5-3.51 (Exam 70-53)".

Just passed TCP/IP 3.5/3.51 today with 857.. here's my 2 cents.

I used the Resource Kit, TCP/IP Help file (3.5 plus 3.51 addition), the Microsoft articles in the Exam description, and the New Riders TCP/IP|SMS book. (I also read Comer volume 1, but I think it was too advanced for the exam.) Also there are plenty of Web resources (I loved the subnet article available from 3Com [even though it has a lot of new material that is not relevant to 3.5/3.51]); for references check out any of the personal Web sites related to MCSE exams. I also printed out various articles on TCP/IP from other Web sites.

I wrote down the following charts during the practice exam:
(1) Subnet chart

(2) IP Class net chart (number of hosts/nets per class, starting ending ranges, etc.)

(3) IP-host name resolution order

(4) NetBios name resolution order

(4) Important differences between HOSTS/LMOSTS DNS/WINS.

Well, I wish I could say otherwise but it was not as grueling as I had feared. Sure, the screen is too small with the large question and the complex diagram (and some questions have two exhibits), there's the usual Microsoft ambiguous questions or three. I agree: as has been said before, the main issue is time management. I finished with some 20 minutes to spare; so I reviewed the 5-6 questions I was unclear about, and just got out of there.

I made some comments on test preparation in the thread on the Networking Essentials exam, and these still hold: I wrote out the subnet chart (and the other lists I wanted to remember) over and over on blank pieces of paper until I got it down. I timed myself to be sure I could do it in less than the time of the practice exam (15 minutes). I made flash cards out of index cards about the TCP/IP command line utilities and their switches. I used the command line utilities as much as I could. I used all the demo exams available for download from the various vendors and Microsoft's assessment exam. Tried to do all the subnetting questions I could find for practice.

Some comments on the New Riders TCP/IP|SMS book: it is quite good in that it is another perspective on it than the official Microsoft documentation, particularly things Microsoft is [what appears to be] unclear. Helpful are hints on what doesn't work or what to avoid. Also very helpful are the chapters on troubleshooting, tuning/SNMP. However, the book is *VERY* poorly edited; there are mistakes and some of the graphics are horrid. This book also badly needs a second edition. The CD with it is almost useless. If another publisher comes out with something better, this one is history.

Some important points to remember:

1. IP address/host name resolution: DNS / HOSTS. file. Static. Names are CASE SENSITIVE. # is ONLY uUsed for comment delimited in HOSTS. file

2. NetBios name resolution: WINS/ LHOSTS. file. WINS is Dynamic. Names are UPPER CASED automatically. # is used for comments and certain special flags; ie: #PRE, #DOM, #INCLUDE, etc.

More on the Exam....

1. Resource Kit refers to the NT 3.5 Resource Kit Vol 2: "Windows NT Networking Guide".

2. Here are the mnemonics I used for IP-host name resolution order, and NetBios name resolution order [slightly modified from a previous post]:

a. NetBios names: "No Way Baby Let's Have Dinner": NetBios Name cache, WINS, Broadcast, LMHOSTS., HOSTS., DNS.

b. IP-host name resolution: "Let's Have Dinner When Baby Leaves": Local host, HOSTS., DNS, WINS, Broadcast, LMHOSTS.

3. Also know the difference between the four NetBios node types and how to use the value that determines each (1 = b-node, 2 = p-node, 4 =m-node, 8 = h-node).

Gosh, I hate these questions, I am always so paranoid that I will make a typo.

This is especially worrisome because in almost every one of these the test creators screwed up the addresses in the question.


re: <<

1) On routing table entries (route add.....), is a destination entry to a specific host id legitimate or must the entry be to the network address. For example, can the command be: route add 192.100.1.1 mask 255.255.255.0 131.107.25.25. Or, must it be to the network address 192.100.1.0? I know the latter is more "global", but I am interested in this as a point of information.

By the way, is the command case-sensitive?

 

Yes, you can route to a single address. If you do route print, you will see examples of this added by the 'default' route configuration. Your own box will be in there as will the default gateway. These are individual stations. The mask for a specific station is: 255.255.255.255, meaning all bits must match.

Well, you must use individual addresses when that is what your are targeting and network addresses when you want to route to a group of machines.

Usually you add entire NETWORKs for those outside your network, entire subnets for those not on your subnet, and individual stations when there are special considerations or problems. No one wants to add a bunch of these when one entry will solve the problem.Again compare the route for your OWN machine to the route for its network. Their are two DIFFERENT entries.


re: <<

Also, what about the case where a network is subnetted. Transcender poses the following type of question: From host 131.107.45.29 you need to connect to host 131.107.90.198. The local gateway address is 131.107.45.1. The subnet mask is 255.255.248.0. What is the routing table entry? Their answer indicates that the answer would be: route add 131.107.90.0 mask 255.255.248.0 131.107.45.1. I know that the network id for this host is 131.107.80.0. Shouldn't 131.107.80.0 be the entry here instead of 131.107.90.0? Will it work the way they have recommended?

 

route add 131.107.90.198 mask 255.255.255.255 131.107.45.1 (will route to that individual host)

route add 131.107.88.0 mask 255.255.248.0 131.107.45.1 (will route to that subnet assuming the mask given -- the '8-bit' is part of the network so it is '88' not '80'. And this is host "2.198"

But you're right, it is definitely NOT 131.107.90.0. The question iseither a typo/oversight or the author didn't understand subnetting thoroughly (or you misquoted <grin).

[Theoretical methods. Pragmatics actually]

These are easy to reason out if you know my 'high-bit sums' trick. 248 is the 5-bit sum, leaving 3-bits in that octet. The 5-bit sum includes the 128-bit, 64-bit, 32-bit, 16-bit, & 8-bit. Of these, the 64, 16, and 8 (=88) are set in the subnet, and with the remaining 3-bits, you can make any number from 1-7.

So each of the subnets available with this mask supports hosts numbered from:

0.1 to 7.254 (never 0.0 or 7.255)

Please note on while we are analyzing this that the following are valid host ids:

1.0 2.0 etc.

0.1 0.2 etc.

1.255 ... 6.255 are all valid

Many people halfway learn the rule, "All of the host bits cannot be either 0 or 1".

Thinking it means no 'octet' can be all zeroes is incorrect. In this case the host id is constructed from 3-bits in the third octet, and the 8-bits of the fourth octet.

My method does NOT depend on memorizing a lot of tables. It depends on learning the basic binary arithmetic tables, including the 'high-bit sums'. You have about 2-3 small tables to memorize and these are easier than learning the multiplication tables.

The really cool part is that if you forget them, they are trivial to generate. You will know how to do these the rest of you life. Every student of mine has scored 100% on the subnetting section.


re: <<

2)Based on Transcender's answers, I'm having trouble coming up with a rule-of-thumb on the "meets the requirements ..." questions. In other words, does anyone have any "theoretical" way to approach these nightmare questions. The cases where the solution either works well or doesn't work at all are clear to me. However, the grey areas are the problem.

Nope. Lot's of people have advice but nothing concrete. I have been led to believe by many that these tend to be A when they are technically correct. Another pundit suggests they are A when they sound like marketing copy.

Most indicate that "Does not meet the requirements/doesn't work" is a frequent alternative. Some suggest these are the only two choices but I wouldn't count on it.


re: <<

3) On the TCP/IP exam, I understand there are questions where you have to correctly enter a command line. Can those experienced with the test please give me an idea on which commands I should be extremely familiar with. I've heard the ROUTE ADD command is common. Any others?

Route add is a big one. Don't expect detailed questions on most of the others. Usually Route add is just Destination & Gateway, no mask or metric. How can you leave out the mask?

The mask defaults to 255.255.255.255, so works correctly for individual hosts.

ipconfig: /release /renew /all You may not get questions but you should know what they do, you will get a question perhaps where that is important and these are VERY useful switches that many people overlook.

Is it case sensitive? Please go type one (you can always delete it afterward.) [Though I forgot, didn't you?]

I passed my TCP/IP 3.51 exam recently. The following is what I recalled:

* Subnet mask is very easy. Many books advised you to memorize the table to save time. I think it is not neccessary.

* Some questions are about UNIX printing. It involves HP's DLC protocol. If you install the protocol and driver in your server, there are very few configurations on the client side. Elminiate the options about doing a lot of configuration in UNIX.

* The questions on DHCP are lengthy and the graphs would confuse you. Remember: DHCP only overrides IP and subnet mask. It does nothing to your locally configured default gateway and WINS. Don't panic if you see a very complicated graph and screen shot.

* Overall questions are tricky and graphs tend to confuse you. Ignore unrelated information on the graph. For example, one question is about the number of required networks. The graph shows ethernet and token ring. The topology doesn't matter!

Hope it helps. Thanks.


I passed TCP/IP NT 4.0 test #70-059 on Sunday with a 913 which means wether Microsoft is aware of it or not, there is another MCSE on the list.

My track to achieve MCSE status was

Networking Essentials #70-058

NT Workstation 4.0 #70-073

NT Server 4.0 #70-067

NT Server in the Enterprise #70-068

IIS 3.0/Index server 1.1 #70-077

TCP/IP on NT 4.0 #70-059

 

I feel pretty good about my performance overall on the six test I took since my average score was 910. I plan on sticking around this listserve awhile since I am planning on taking some more test. maybe go for that MCSEx2 when NT 5.0 rolls around.<Grin

As far as a brain dump on the previous test.. I have already shared all I am going to on those test! As for the TCP/IP test I took today. the only help I followed and would recommend to others is an excellent article on the beta version of the test located at.

http://www.mcpmag.com/members/97mayjun/col2main.asp

Along with a TCP/IP subnetmasking chart and article located at.

http://www.mcpmag.com/members/97mayjun/oe1main.asp

I found that the whole chart all three sections was worth memorizing since it helped me breeze through the test. These charts may seem hard to memorize at first but the numbers are just powers of two -2 for the most part.

In addition as study material I read the following:

Mastering Windows NT Server 4 ISBN#0-7821-2067-9 Chapters 14,15,17 and Appendix A

Windows NT Server 4 Professional Reference ISBN#1-56205-731-6 Chapters 14-18 and 23

I also read the complete Sybex TCP/IP Study guide ISBN#0-7821-1969-7 cover to cover, although I do not recommend that anyone use this book for the test since the 4.0 updated material amounted to just one small poorly done chapter in the back of the book.

As additional reference material I cross checked several items against the Windows NT Resource kit and the Microsoft Technet CD's.

That's all there is to it.... Lots of reading,cross checking and hands on trying stuff out is all it takes to achieve high scores on the test in my opinion.

2 hours, 58 questions 750 required.

passed 879

planning 85 / installation 96 / Connectivity 80 / Monitoring and =

optimisation 100 / Troubleshooting 73


Gosh, this was quite hard, I would recommend skipping the first 40 odd questions and getting to the trouble shooting section first, I was quite mentally tired by the time I got here and these questions require clear thinking and my sections scores reflected that. Lots of questions (8) in the form " You can access service / machine by name/address using utility x/y/z, but not by address/name using utility x/y/z what is likely problem.

Other stuff.lots of giveaways (10) on what utility arp / netstat / nbtstat / ipconfig / netmonitor / perfmonitor / snmp does what. Didn't even ask with what options, could be that's fair enough, so long as you know which utility you can always lookup the options. 5 questions on subnetting. 2 questions on tcp ip printing one unix - nt one nt - unix.

1 lmhosts, one hosts, one replication, 4 situational (does that word exist?) questions on multihoming with dhcp and snmp. 5 Situational questions with IP routing (both static and dynamic) and browsing and ftp in a heteregenous environment.

Couple of dns questions, should you install a forwarder/cache server/secondary

Many questions involved, sometimes peripherally, wins and dhcp, know them!

what options apply at what level for dhcp , what info must be supplied when reserving ip addresses.

Overall I was a bit disappointed with this test??? I don't think it goes into sufficient detail, e.g why would you use ... instead of what would you use but there you go, that's just my opinion.

Well that's enough from me, any incoherence can be blamed on Jamesons and Stella (not in the same glass, that's just ugly.) If any one has any specific questions fell free to e-mail. I shall remain on this list to give my opinions on questions raised so that others may benefit just as I have!. Nicely put huh?