two questions about switch

pengbin

hi, I am new to CCNA and have two questions about switch.

1. Read one CCNA book. It says "The main purpose of a switch is to make a LAN work better-to optimize its performance-provide more bandwidth for hte LAN's users." As I know, each port in the switch breaks up collision domains but can't break up broadcast domains. So it can reduce the collision in the LAN. But can't understand how it can provide more bandwidth? anybody explain?


2. Can switch be connected to directly to Bridge or hub?

BubbaJ

1. You can run full duplex with a switch. A switch's backbone has a lot more bandwidth than any individual port. A switch can have an interface with more bandwidth to the rest of the network.

2. Yes. You will probably need a cross-over cable.

Danman32

1. Another reason a switch can increase throughput is because multiple frames that have uniqe source/desination macs from other frames can be forwarded at the same time. If a host on port A wants to send a frame to a host on port B, it can, even while a host on port C is sending a frame to port D.
Also, if there is only a single host on a switch's ports, then collsions are elliminated, and the full bandwidth of the cable can be utilized.
Otherwise, because of the nature of ethernet being CSCD, statistically as more hosts try to fully utilize the ethernet, the usefull bandwidth goes down as collisions rise. Statistically it can be shown that the full utilization in such a case is about 1/3 the cable capacity.
Think of a multi-way conference call, where only one person can talk at any one time, but anyone can start saying something. If everyone had a lot to say, it could be hard for everyone to get their thoughts out as they are likely to collide with someone else trying to talk. Compare this with calling and talking to only one other person. Both of you could converse, using the entire bandwidth of the line with no collision.

2. If the hub or bridge had an uplink port, or the switch had an uplink port, or if either has an auto-sense, you may not need the crossover. Most switches today have auto-sense.

binarysoul

I'm very confused about switching although I've studied well (using Odom's book and Sybex). Please help.

1. Does a switch only have 2 interfaces, one for incoming and second for outgoing?

2. When a device sends a frame, it puts the destination address of another device, but how does the frame gets to the switch since the frame is NOT addressed to the switch?

3. Can the destination address of frame also be added in the MAC table? I read that only source address is added to MAC table on the switch as the switch learns. How long does each MAC entry reside in the table?

Webmaster

Try this one for a basic explanation of switches:
www.techexams.net/technotes/networkplus/networkcomponents.shtml

I'm very confused about switching although I've studied well

If you studied switching 'well' you should at least know what a switch looks like. So in regards to your first question, check out the following pages:

www.cisco.com/en/US/products/hw/switches/ps628/prod_view_selector.html
www.cisco.com/go/2950/
www.cisco.com/en/US/products/hw/switches/index.html

mwgood

pengbin wrote:

As I know, each port in the switch breaks up collision domains but can't break up broadcast domains. So it can reduce the collision in the LAN. But can't understand how it can provide more bandwidth? anybody explain?

Once you really understand what a collision domain is and how it operates, then you should see how using a switch appropriately can increase bandwidth. See the link Webmaster posted for this.

The fewer the number of people who try to walk through the same door at the same time - the better.

Also - the option of full duplex will increase bandwidth as well.

rakem

binarysoul wrote:

I'm very confused about switching although I've studied well (using Odom's book and Sybex). Please help.

1. Does a switch only have 2 interfaces, one for incoming and second for outgoing?

2. When a device sends a frame, it puts the destination address of another device, but how does the frame gets to the switch since the frame is NOT addressed to the switch?

3. Can the destination address of frame also be added in the MAC table? I read that only source address is added to MAC table on the switch as the switch learns. How long does each MAC entry reside in the table?

1. Switches have muliple ports, anywhere from 8 up to 96 or even higher. All ports are capable of sending and receiving, unless they are blocked or disabled by either the administrator or something like Spanning Tree.

2. The frame gets to the switch because the computer sending the packet is connected directly to the switch. It has to hit the switch first to learn where it is going.

3. It can vary, if the switch hasent received a packet from a certain MAC address it will flush the MAC table. not to sure on the exact timers thou.


oh and when your talking about switches (layer 3 devices) its packets. Frame belong to routers or layer 2 devices.

binarysoul

Thanks everybody for the answers. Here is an interesting question

In which of the following CLI modes could you issue a command to erase the switch's initial configuration?

a. User mode
b. Enable mode
c. Setup mode
d. Global configuration
e. Interface configuration mode

The correct answer is B, but what about choice D, Global configuration?

Global configuation is also an 'enabled' mode that affects overall router operation. Why would we not accept d as an answer?

BubbaJ

binarysoul wrote:

Why would we not accept d as an answer?

This is probably not covered for the CCNA exam, but, in the configuration modes, current IOS versions (I think 12.3 and 12.4) allow you to use the do command in front of a priviledged exec (enable mode) command. You could type do write erase.

Danman32

rakem wrote:

oh and when your talking about switches (layer 3 devices) its packets. Frame belong to routers or layer 2 devices.

You've got that backwards.

layer 1, physical: the ethernet cable itself. Also part of the NIC hardware
Layer 2, data link: hubs and switches that move ethernet frames.
Layer 3, network: This is where the IP is addressed and IP packets form. Routers forward IP packets from one network to another when the destination network is different from the source.

Hubs are like the party line phones of yesteryear. Everybody got the conversation to their phone and could listen in if they wanted to. If someone else was using the line, you couldn't until the line was clear. Begging, pleading and bribery often helped.

Switches are like the phone system of today. You have a phone number, your friend has a phone number. You dial your friend's number with your phone, and the phone company's switch would connect your line to your friend's line. With an ethernet switch, the switch looks at the destination address of the frame you sent, and if it knows which port is connected to the destination host, it will forward that frame to that port only. While it is at it, if it doesn't know the port of the soure address in the frame, the switch notes what port the frame came in on and maps that port to that source MAC address.

You MUST KNOW THIS WELL to pass the test.

unwritt3n

Danman32 wrote:

Layer 2, data link: hubs and switches that move ethernet frames.

A hub is a layer 1 device.

BubbaJ

unwritt3n wrote:

A hub is a layer 1 device.

Yes, it is.

Danman32

unwritt3n wrote:

Danman32 wrote:

Layer 2, data link: hubs and switches that move ethernet frames.

A hub is a layer 1 device.

That is true. I tend to look at it as layer 2 though, since it is comparable and a direct replacement component to a switch. But I do realize and understand that industry and exams consider it layer 1 and why they do.

BubbaJ

Danman32 wrote:

I tend to look at it as layer 2 though, since it is comparable and a direct replacement component to a switch.

It is in the sense that you use it to provide network ports, but it is also much like cabling (Layer 1) in that respect.

On the other hand, it in no way is comaparable or a direct replacement component for a switch since a switch will read the frame headers and make forwarding decisions based on them. I deal with many networks that would just die if a hub were put in place of one of the switches.

binarysoul

I was browsing Cisco's products site to learn about switches and came across 4500 switche which has slots (Modular 3- and 6-slot single supervisor chassis and 7- and 10-slot ).

-What is a slot used for and what is a supervisor chassis? Are these areas we need to know for CCNA?

-Which switch types are covered under CCNA, e.g. 1900, 2950, 3xxx?

BubbaJ

binarysoul wrote:

-What is a slot used for and what is a supervisor chassis? Are these areas we need to know for CCNA?

No, although you may need to know that there are modular switches.

The 4500s are modular to take several different type of boards (100Base-TX, 1000Base-T, 100Base-FX, 1000Base-SX(LX and others), etc. in different densities. IOS runs on the supervisor module, and it can be a multilayer switch that runs routing protocols (depending on the IOS version).