- Posts: 102
- Thank you received: 0
Flooding incase of Switches..
19 years 3 months ago #9349
by tiamat
Replied by tiamat on topic Re: Flooding incase of Switches..
while UDP does not require ACKs, the first thing the switch is going to do is to ARP for the destination machine, at which point host B will respond. At that point, the switch adds host B to it's cam table, and we're done.
- Datacom_guy
- Topic Author
- Offline
- New Member
Less
More
- Posts: 13
- Thank you received: 0
19 years 3 months ago #9373
by Datacom_guy
Replied by Datacom_guy on topic Re: Flooding incase of Switches..
Tiamat,
As the switch will flood the frame, it will not broadcast an ARP request, so the destination node will not send an ARP reply. So we have to wait till the destination node sends a frame!!
As the switch will flood the frame, it will not broadcast an ARP request, so the destination node will not send an ARP reply. So we have to wait till the destination node sends a frame!!
19 years 3 months ago #9380
by tiamat
Replied by tiamat on topic Re: Flooding incase of Switches..
sorry, i always seem to be stuck in 'layer 3 switching' mode... what I should have said was, the first thing host A would do is an ARP request (broadcast) which (if on the same subnet/vlan) the switch will forward and when host B responds the switch will put that in it's table.
Like cybersorcerer said, though, it's not a typical situation where there are lots of frames with an unknown destination on the network (for legitimate reasons anyway). I'm probably too used to cisco equipment with adjustable CAM and ARP table timers to think that switch flooding would be a common problem, but obviously it can be given the proper circumstances.
Like cybersorcerer said, though, it's not a typical situation where there are lots of frames with an unknown destination on the network (for legitimate reasons anyway). I'm probably too used to cisco equipment with adjustable CAM and ARP table timers to think that switch flooding would be a common problem, but obviously it can be given the proper circumstances.
- Datacom_guy
- Topic Author
- Offline
- New Member
Less
More
- Posts: 13
- Thank you received: 0
19 years 3 months ago #9386
by Datacom_guy
Replied by Datacom_guy on topic Re: Flooding incase of Switches..
I also agree what cybersorcer said "ethernet was not designed to provide acknowledgement and leaves those sort of functions to the upper layer protocols. In most cases, "computer b" will eventually respond to a request or request data long before a broadcast storm becomes a problem."
But just felt like if the destination node is specifically made to reply when it receives a broadcast frame (so that the switch learns) the performance can be increased.
But just felt like if the destination node is specifically made to reply when it receives a broadcast frame (so that the switch learns) the performance can be increased.
19 years 3 months ago #9392
by tiamat
Replied by tiamat on topic Re: Flooding incase of Switches..
except I dunno how much of a performance boost you'd get if every host were to respond to every broadcast on the network at the same time. I think that alone would cause problems for the switches.
19 years 3 months ago #9433
by TheBishop
This is a valid and interesting discussion! However I agree that in practice the unknown computer will have sent numerous frames to the switch long before the flooding issue becomes significant. Just stick a hub on the end of a windows PC, plug in another PC running a packet capture utility and switch the windows PC on. You'll find a veritable cornucopia of frames sent out as the machine starts up, tries to find domain controllers, DNS servers, WINS servers etc etc
Time to create page: 0.132 seconds