Few basic questions

Hi everybody,

I have a few simple questions that I'm sure I can get answears to here.

1) Isn't enough for one end port of a TP cable to be Auto-MDI for it not to matter wether the cable is ST or CO?

2) I know that the max. number of nodes for a 10BaseT network is 1024, but is that per hub, per collision domain, per broadcast domain or?

3) I'm reading this book "Network+ Training Guide" and one end-chapter question is:
"You want to create a larger network by connecting two switches together. One of the switches has a port that can be switched from MDI to MDI-X as needed. The other switch doesn't have such a port or a dedicated uplink port. Which type of cable should you use, and how should you configure the switchable port to create the larger network?
a. Use a ST cable and set the port to MDI
b. Use a CO cable and set the port to MDI-X
c. Use a ST cable and set the port to MDI-X
d. Use a CO cable and set the port to MDI"
I marked answear a and b to be correct, but the book says only answear b is correct.
I'm think the book is wrong, wouldn't be the first time

Cheers, MezzUp

]Let's have a first stab then:

1) Isn't enough for one end port of a TP cable to be Auto-MDI for it not to matter wether the cable is ST or CO?
At its simplest, think of a UTP cable as having a transmit pair and a recieve pair in it. SO when connected between two devices the transmit of one device needs to be wired to the recieve of the other and vice versa. When you connect a hub port to a PC, this 'crossover' is taken care of by the way the hub port is wired so you can use a straight-through cable. However to connect a hub to a hub (or switch) you would then get a double-crossover which wouldn't work. So you have to use a crossover cable to get the correct connection.
So (we get there eventually) if one of your devices does auto-crossover then whatever cable you plug in should work since the port will put itself into the correct mode

2) I know that the max. number of nodes for a 10BaseT network is 1024, but is that per hub, per collision domain, per broadcast domain or?
Eh? Wot? Don't know anything about this one...

3) "You want to create a larger network by connecting two switches together. One of the switches has a port that can be switched from MDI to MDI-X as needed. The other switch doesn't have such a port or a dedicated uplink port. Which type of cable should you use, and how should you configure the switchable port to create the larger network?
a. Use a ST cable and set the port to MDI
b. Use a CO cable and set the port to MDI-X
c. Use a ST cable and set the port to MDI-X
d. Use a CO cable and set the port to MDI"
I marked answear a and b to be correct, but the book says only answear b is correct.
I'm think the book is wrong, wouldn't be the first time
The MDI (medium dependent interface) is the component of the media attachment unit (MAU) that provides the physical and electrical connection to the cabling medium. MDI ports connect to MDIX ports via straight-through twisted pair cabling (see my hub to PC example in 1) above. But if you want to connect both MDI-to-MDI or MDIX-to-MDIX connections you must use a crossover cable.
So, the switch with no MDI/MDIX facility will have an inherent MDIX configuration on its port. If we want to connect to an MDI port then we need a straight-through cable so a) should work. And if we want to connect to an MDIX port then we need a crossover cable so b) should work too. By the same logic c) and d) won't work. The book does seem to be wrong doesn't it? However this might be one of those "we don't want the correct answer but the best answer" scenarios beloved of those slippery souls who compile certification exams. But I still can't see why only b) would be 'correct'.
What does everyone else think?

To answer question number 2, I think he was reffering to maximum amount of nodes that can contend on an ethernet in just one collision domain without suffering performance degradation. This is just the 802.3 reccomendation for ethernet design. But once you throw switches in the mix, this counter is reset with every seperate collision domain that you have. Thus, this limitation is basically a problem of the past. I still don't know why these tests demand you to know obsolete technology and trivial facts... It's like memorizing IRQ's for the A+. Haven't those comptia dudes heard of plug n' play???

To be honest I wanted to reply here yesterday, but the last question left me uncertain and so I waited in hopes someone more experienced in ethernet (picturing Chris and Bishop) would answer

Anyway, A and B seem reasonable to me too. It is all about having an odd number of reversals after all, so that Tx connects to Rx and vice versa, right? If so, A and B should be at least equally correct! Unless we all miss something

The answer to this exact problem, from informit.com, here , is:

Because one of the switches does not have MDI capability, the switchable port should be set to MDI-X. Then, a crossover cable should be used to cancel out the crossing between the two devices. None of the other options would result in a successful connection. For more information, see the section "Working with Hubs and Switches," in this chapter.

Still, I could not find any reason suggesting this, in their refering article. Furthermore, the way I see it, if we use both ports in MDI-X mode, then there is no "crossing between the two devices that needs to be canceled".. Tx and Rx pairs would be reversed twice, so they will maintain their original aligment, and that's why we would need a CO cable to reverse it.

Funny thing is the two backbone switches at my network are connected according to option A, but they also support Auto-MDI -as all of today's switches I know, even the cheapest ones-, so that is not a proof.

PS. Cybersorcerer is right on 2, of course if you use hubs to connect so many nodes, you'll have a huge collission problem much sooner before you hit 1024!

Thanks everybody that have replied.

Another quote from the book: "Roughly translated, 1 Gbps is 1 MBps. That means it would take 10 second to transfer a 10 MB file across a link...."
It supposed to be 125MBps, right?

These errors are forgivable, if I could find an errata online, but I've looked alot without success. An ISBN lookup at Que Publishing website turns up empty. If anybody wants to give it a try the ISBN is 0-7897-2830-3.

I do also have one more question. This is a quote from the book, which does make sense, but I thought(and had read) that is was just a marketing gimmick.
"On a network, network card that can use full-duplexing can double their transfer rates. For instance, a 100Mbps network card in full-duplex mode can operate at 200Mbps"
Is this true or false?

Gbps is 1 MBps. That means it would take 10 second to transfer a 10 MB file across a link...."
It supposed to be 125MBps, right?

yes, 1Gbps / 8 = 125MBps minus the overhead of the associated protocols and any other caused by less-than-ideal conditions that exist in the real world

On a network, network card that can use full-duplexing can double their transfer rates. For instance, a 100Mbps network card in full-duplex mode can operate at 200Mbps"
Is this true or false?

Well, phrased like this it sounds like a marketing gimmick.. Full duplex allows transmissions at both ends to utilize the full width of ethernet at the same time, thus in the case of fast ethernet, allowing an 100mbps upstream and an 100mbps downstream simultaneously