2724 uplink?

Question

i would like to 'chain' together four 2724's(A,B,C,D) can i maintin a full gigabit connection from 1 to the next simply by plugging in port 24 on A to 23 on B and 24 on B to 23 on C and 24 on C to 24 on D?

GregG1 · Answer

This configuration would provide connectivity, but is not recommended. Daisy-chaining multiple switches is not an efficient topology. It is not a scalable solution and can cause performance problems. A more efficient and scalable solution is the star topology in which one switch acts as the 'core' and all switches connected directly to the core.

GregG1 · Answer

There is no dedicated uplink port on the switch. All ports function the same.   Correct, this would be a more efficient topology.   The 2700 series does not support Spanning Tree. If all switches are connected directly to the 'core' switch, however, there should be no redundant loops for Spanning Tree to block.

scottadmi · Answer

3 questions:

1. so i should connect all switches via any port to switch A?

2. is there a specific uplink port if your not using fiber?

3. will this create alot of broadcast traffic or will spanning tree take care of it?

scottadmi · Answer

so then we would have:

Switch A serving as the core switch.

Switch B port 1 connected to Switch A port 1

Switch C port 1 connected to Switch A port 2

Switch D port 1 connected to Switch A port 3.

and this is the recommend setup, correct?

Thanks so much for your input it is greatly appreciated.

scottadmi · Answer

if the switch does not support spanning tree then how does it prevent looping and limit broadcasting?   does this mean that i should by a dell switch which does support spanning tree and use that as the core switch?  will that prevent looping and excessive broadcasting?

scottadmi · Answer

actually i just re-read all our posts and my last post please disregard the looping question you answered that.

my main concern now is broadcast traffic. here is my thought:

you have two 2724 switches A and B. you also have 2 hosts 1 on each switch.

host 1 on switch A.

host 2 on switch B.

----------------------------------------------

host one wants to transmit a packet to host 2. so switch A looks at its mac address table and does not see host 2's mac and therefore sends it out on all ports. port 10 on switch B is connected to port 10 on switch B. When switch B recieves the packet it has the mac address for host 2 and switches the packet directly to it. in reverse the same must happen only switch B now must broadcast to all ports and switch A has the mapping so it does just a packet switch.

doesn't this create ALOT of broadcast traffic or is there something that prevents this from happening over and over.

sorry for the long post but i hope you can see what i am asking

cheers.

GregG1 · Answer

What you are describing is the flooding of unknown unicasts that occurs when the switch has not learned a MAC address. This is a necessary step for Layer2 transparent bridging and cannot be helped. The bridge aging timer on the switch should prevent this from happening too often.

If you see excessive unknown unicasts on your network, you may need to look into it. One possible source would be a "flapping" port. The switch flushed MAC addresses on a port when the port goes down. A flapping port (rapidly transitioning up/down) would continuously flush the MAC addresses on the port.

scottadmi · Answer

in this scenario am i broadcasting everything that i want to transmit back and forth from host 1 to host 2

scottadmi · Answer

...assuming all hardware is working properly, do you know if the broadcasts which i described will always occur.  Will switch A always have to broadcast on all ports in order to get the packet to go to switch B and vice versa.  Thats my question i am not having an issue i am trying to research to find a solution before implementation.

scottadmi · Answer

...again i wanted to say thanks for your input it is very much appreciated.

scottadmi · Answer

last question and its a big one. do the 2724's support transparent bridging.

scottadmi · Answer

dood you ROCK!!! Thanks.

GregG1 · Answer

When client1 wants to talk to client2, the following happens (assuming the clients are using static IP addresses and do not broadcast anything when initially connected):

Client1 sends an ARP request to determine the MAC address for the IP address used by client2
The ARP request is sent with client1's MAC address (source) and FF:FF:FF:FF:FF:FF (destination)
As the traffic enters switchA, the source MAC address is learned on the incoming port. SwitchA floods the frame on all other ports (except the incoming port)
The same procedure happens for each switch the traffic passes. Each switch learns the source MAC address on the incoming port and floods the frame on all other ports.
When the ARP request reached client2, it responds using a unicast with client1 MAC address (destination) and it's own MAC address (source).
As the traffic enters switchB, the source MAC address is learned on the incoming port. As the switch has already learned the destination MAC, it knows which port to forward the traffic.
The same procedure happens for each switch the traffic passes. Each switch learns the source MAC address on the incoming port and forwards the frame to the port on which the destination MAC was learned

For more information on transparent bridging, see the following link:

http://computer.howstuffworks.com/lan-switch10.htm

GregG1 · Answer

All PowerConnect switches (as well as most other vendors) forward Layer2 traffic based on transparent bridging practices.

Networking General

2724 uplink?

Was this post helpful?