A couple of questions about... questions.

blackmage439

I ran into a couple of ambiguous questions while studying my Net+ book.


What would be the effect on your overtaxed network if you reprogrammed all client and server Application layer protocols to use UDP rather than TCP?

a. Due to UDP's lack of error checking, traffic would travel somewhat faster and network performance would improve.

b. Due to UDP's connectionless state, no bandwidth would be wasted on sequencing and flow control, and network performance would dramatically improve.

c. Due to UDP’s unreliability and the amount of traffic on the network, more errors would likely occur and traffic levels would increase, resulting in slower response time.

d. Using UDP rather than TCP would have no effect on the network's performance.

I chose C. It just seems logical since TCP is in place for a reason...


And


19. You manage a server that allows university students to use Telnet to make a connection, then initiate a FTP session to upload their homework. Professors also pick up students' homework by telnetting to the computer and using FTP You have decided to change the FTP port number on the server from its default to 23, for better security. Assuming students and professors make no changes to their default workstation configurations, what will be the result of this change?

a. Students and professors will be able to Telnet to the server, but unable to FTP files to and from the server.

b. Students will be able to Telnet to the server, but professors will be unable to do either.

c. Students and professors will be unable to Telnet to the server, but able to FTP files to and from the server.

d. Students and professors will be unable to Telnet to the server, or FTP files to or from the server.

I chose D. Since Telnet (default port 23) and FTP requests will conflict with eachother, neither should work.

snadam

these questions look awfully familiar. are you using the thompson book?

blackmage439

Haha. Yes I am. I don't like it as much as their A+ book, but it still is pretty good.

mgmguy1

blackmage439 wrote:

I ran into a couple of ambiguous questions while studying my Net+ book.


You manage a server that allows university students to use Telnet to make a connection, then initiate a FTP session to upload their homework. Professors also pick up students' homework by telnetting to the computer and using FTP You have decided to change the FTP port number on the server from its default to 23, for better security. Assuming students and professors make no changes to their default workstation configurations, what will be the result of this change?

a. Students and professors will be able to Telnet to the server, but unable to FTP files to and from the server.

b. Students will be able to Telnet to the server, but professors will be unable to do either.

c. Students and professors will be unable to Telnet to the server, but able to FTP files to and from the server.

d. Students and professors will be unable to Telnet to the server, or FTP files to or from the server.

I chose D. Since Telnet (default port 23) and FTP requests will conflict with eachother, neither should work.

My question is, instead of changing port numbers?,Would it not make more sense to use
Secure File Transport Protocol - SFTP since you gain the benefits of public key encryption and compression?

Slowhand

mgmguy1 wrote:

My question is, instead of changing port numbers?,Would it not make more sense to use Secure File Transport Protocol - SFTP since you gain the benefits of public key encryption and compression?

It probably would, but when you're doing these types of questions, you have to work with what they give you. The solution they're looking for isn't always the solution you'd implement, yourself, if you were doing it on your own.

Slowhand

And in response to blackmage439's original post. . . for the first question, I would go with A, and possibly B. I wouldn't pick C, mainly because UDP isn't called "unreliable" because it causes a lot of errors. It's unreliable because it doesn't check for errors and it doesn't pre-emptively check for the availability of the receiving host. It just sends off its packets, and goes on about its business. TCP, on the other hand, spends some time and overhead setting up a connection, sequencing data, and then waits for a response. It also takes up some bandwidth sending re-transmitted data that wasn't accounted for by the receiving host. All of this error-correction and checking takes up overhead. If you were to replace all your TCP apps with ones that use UDP, (and most UDP-based apps tend to have their own error-checking, higher up in the TCP/IP stack,) then the result would be an increase in network speed because of the lower overhead cost of connection-less communication.

The second question, I'd tend to agree with you on. Depending on the specific FTP and Telnet applications, there might be some warning messages or some accomodation for the clash. . . but I'd imagine that, in general, the two apps would be in conflict over the port and wouldn't be able to function.

blackmage439

Slowhand wrote:

And in response to blackmage439's original post. . . for the first question, I would go with A, and possibly B. I wouldn't pick C, mainly because UDP isn't called "unreliable" because it causes a lot of errors. It's unreliable because it doesn't check for errors and it doesn't pre-emptively check for the availability of the receiving host. It just sends off its packets, and goes on about its business. TCP, on the other hand, spends some time and overhead setting up a connection, sequencing data, and then waits for a response. It also takes up some bandwidth sending re-transmitted data that wasn't accounted for by the receiving host. All of this error-correction and checking takes up overhead. If you were to replace all your TCP apps with ones that use UDP, (and most UDP-based apps tend to have their own error-checking, higher up in the TCP/IP stack,) then the result would be an increase in network speed because of the lower overhead cost of connection-less communication.

At first I chose A, but then I started thinking in terms of networking.

UDP is considered unreliable exactly for the reasons you described (no error checking, connectionless), and it is great for the reasons you said (faster data transmission). However, the bad make it an even bigger problem on over-taxed networks.

*The following is rambling for those who are not familiar with networking hardware. *

Hubs are inherently faster than switches. They do not waste time analyzing data packets. As such, they are a perfect choice for smaller, low-traffic networks. However, once the network increases beyond a certain scope, the use of hubs becomes counter-productive. Erroneous packets caused by packet collisions and runts (a packet missing some of its information), coupled the the needed re-transmission of data, would bring the network to a crawl. The solution is to implement switches. Switches (usually) store, scan, and then forward data packets, thereby eliminating most errors soon after the source.

*End ramble*

In this case, hubs represent UDP, and switches represent TCP. While the hubs are faster at forwarding data, in a high-traffic network they actually add to the problem by inherently creating errors and increasing the traffic load. If it were just a normal, small, non-taxed network, I would definitely agree that UDP would be a viable choice for protocol. However, in a busy network, TCP is by far the better option.

Slowhand

I agree with your comparison of hubs vs. switches, but your original question from your book dealt with programming applications to use UDP instead of TCP. Without taking the network hardware into consideration, you're asked what the result would be. Of course, C would apply as well, and it may be one of the choices. How many answers are you supposed to choose, (just one, two or three, or 'all applicable answers')? I'm thinking, though, that since the question explicitly mentions the OSI model in relation to software apps using specific protocols, I'm thinking that this question deals more with a layer 4 issue, rather than a layer 2 and/or 1.

Of course, I'm sure we could both be completely off-track, and the book will give you a totally different rationalization for the answer they're looking for.

blackmage439

Haha, how true.

In this case, it's only a single-answer question. So, yeah just taking the context of this question, I would agree that A is the correct answer. Considering this is a CompTIA book, I'm not surprised at the question. :P

Thanks for your help, Slowhand!

Slowhand

No problem. Half the battle is figuring out what the problem is, and the other half is figuring out how to read the question and figure out what the test wants for an answer. CompTIA is pretty forgiving, but the vendors, like Cisco and Microsoft, can get pretty finicky.

If you ever go down the Microsoft certification path, you'll learn the old saying, "There's the right way, the wrong way, and One Microsoft Way".