From: ccie2be (ccie2be@nyc.rr.com)
Date: Sun Mar 16 2003 - 10:02:30 GMT-3
Hey Elmer,
Well, it's reassuring to know I'm not the only one having a problem figuring
out this MVR stuff.
I suspect it's not all that important but I much prefer to reach that
conclusion after I really understand it rather than before.
Hopefully, there's someone here in Group Study that can explain this. We'll
see.
Jim
----- Original Message -----
From: "cebuano" <cebu2ccie@cox.net>
To: <ccielab@groupstudy.com>
Sent: Saturday, March 15, 2003 11:05 PM
Subject: RE: 3550 - MVR
> I'm in the same boat. Basically this is one of the many technologies
> better understood if presented in a hands-on type of example, instead of
> just being described. Similar to how Caslow does in his book. Now if
> only there was a practical workbook on the 3550 so we could test out the
> features using TESTED and WORKING samples.
> Sorry Jim, I'll have to re-read this chapter for the nth time before I
> can answer your questions.
>
> Elmer
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> ccie2be
> Sent: Saturday, March 15, 2003 8:21 PM
> To: Group Study
> Subject: 3550 - MVR
>
> Hi all,
>
> I just finished reading for the nth time, the section on MVR (Multicast
> Vlan
> Registration) in Chapter 18 of the 3550 Configuration Guide. I don't
> know if
> it's just me or if that section just isn't all that clear but I really
> don't
> get it.
>
> I have a bunch of questions on it but my first question is this:
>
> How important a topic is this? If I can't get this right will I lose
> all the
> points for Multicast?
>
>
> 1) Regarding MVR itself, it seems that this feature creates a
> separate vlan
> for multicast traffic and that access ports connected to hosts that want
> multicast traffic "belong" to both their normal vlan and the multicast
> vlan.
> But, how can that be if access ports can only belong to 1 vlan? (Ports
> that
> carry the traffic of multiple vlans are trunks.)
>
> 2) Consider the situation where you have a multicast router connected
> via a
> trunk to a Cat switch that doesn't have this feature. On the Cat switch
> there
> are several vlans defined: vlan 5, 6, and 7 with hosts attached to each
> vlan.
> Let's also say, CGMP is running between the router and the switch.
> Here's my
> understanding (taken from the Beau Williamson book) of what happens when
> a
> host wants to join a multicast group.
>
> i) host sends an igmp report to the router (using the mac addr
> equivalent
> of the ip mcast group)
> ii) switch floods mcast frame to all ports regardless of vlan
> iii) the router adds that m-group to its outgoing interface list if
> it's
> not already there.
> iv) router, using cgmp, informs switch to add the m-group to it's
> cam
> table
> v) the switch, using info from the cgmp message adds the host's port
> # to
> the m-group entry
>
> As a result, the cam table will now have 2 mac addresses associated with
> the
> port:
> a) the mac addr of the host hanging off the port used for regular
> unicast
> traffic and
> b) the mac addr of the ip multicast group so that mcast traffic goes to
> that
> host
>
> What's interesting in this process, is that all this activity seems to
> be
> completely independent of vlans. If I'm understanding this correctly,
> unknown
> multicast traffic is flooded to all ports - not just the ports in the
> same
> vlan unlike broadcast traffic which only goes to those ports in the same
> vlan.
>
> So, here's what I can't figure out: How does the above process change
> with
> MVR? And, what's the benefit of MVR over the above process?
>
> I sure hope someone out there in group study can help me with this. I'd
> really appreciate that.
>
> Jim
This archive was generated by hypermail 2.1.4 : Sat Apr 05 2003 - 08:51:40 GMT-3