Question regarding VLAN trunking and hub.

[normanjeter]

This may sound like a wierd question, but a hub, as I understand it, is basically a glorified repeater, right? If you were setting up a trunk between 2 switches (say for example both are 2950XL's), can you not put a hub between the 2 and still have it trunk all the vlan info? I understand trunk ports can only be used between switches (and/or routers), and that access ports are used for any kind of PC attached to it, but why would a hub cause it to only work as an access port? From a network admin's point of view would that not make it easier to sniff out traffic?

Thank You in advance for any info or clarification you can provide.

[Giuseppe Larosa]

Hello Norman,

an hub can only work in half duplex mode for its need to replicate the electrical signal received on port x on all other ports.

This should be the only effect of the introduction of an hub between two switches.

hub works at OSI layer 1 and it doesn't know what is the difference between: an ethernet frame, an ISL frame or an 802.1Q tagged ethernet frame.

Having an half duplex trunk is something to be avoided for performance and stability reasons.

Connecting multiple switches to an hub would be again a performance and a stability problem defeating MAC address filtering capabilty of switches.

So it is not recommended but an hub should not block tagged frames the reasons for not using it between switches with L2 trunk ports are performance and possible instabilities caused by unnecessary replications and reduced peformances caused by half-duplex = single collision domain.

Hope to help

Giuseppe

[lamav]

Hi:

In addition to Giuseppe's observations, take note that using a hub between two trunked switches can increase convergece time considerably. Imagine the following topology:

S1---------S2---------S3

<------------Direction of BPDU flow

S3 is the root

If the link between S3 and S2 fails, S2 will immediately detect the link failure and realize its lost its path to the root. It will then generate an inferior BPDU and send it to S1, which will begin the reconvergence process. With Backbonefast, the reconvergence time can be almost instantaneous.

Now, place a HUB between S3 and S2 and fail the link between S3 and the HUB, maintaining the link between the HUB and S2.

S1---------S2---------HUB----X-----S3

<------------Direction of BPDU flow

S2 will not realize that the link is physically down, but it will stop receiving Hellos from S3. It will nonetheless have to wait for the max_age timer (20 seconds by default - 10 Hello times) to expire before aging out the BPDU it had before the link failure and then claiming itself as the new root and sending out the inferior BPDU to S1.

Hope this didnt confuse you...

Victor

[normanjeter]

To Giuseppe and Victor,

Thank you for your responses on this issue, It was mainly a reference to a mock-up diagram used in a lesson by Testout (the company my workplace uses for preparing people for different Certifications). They currently have me preparing for the first test of my CCNP, and this was one of the mock-up diagrams they used in conjunction to one of their practice questions. The question/answer implied this was not a possibility and the question nagged me in the back of my mind as to why. So technically, while not recommended and highly unstable, it would be possible to have it setup that way, though I wouldn't want to do it at all, lol.

Thanks Bunches!

Norman

P.S. In the unlikely event of ever actually deciding to try this, wouldn't it be possible to, with a setup of this kind, connect a pc to the hub and sniff the trunk traffic, doing full data captures of all allowed vlan traffic on the trunk in question? I know there are better ways of doing it I.E. Replicating a port and using that instead, but this was the scenario I could come up  with for the only possible reason one would consider this type of setup.