I think that Peter Van Oene hit the nail on the head (and confirmed my
conclusion :) , so I thought that I'd share a couple of his thoughts.
" ... More specifically, which applications can work in a unicast
only
world? Do you intend on statically mapping all your IP to MAC
relationships on node by node basis since ARP no longer works as a
discovery mechanism?
Thinking about this stuff leads to the understanding that
broadcasting
is a fundamental communication tool in today's networks and one
cannot
eliminate its use without creating a major disturbance.
Your understanding of VLAN'ing as a very simple technology is on the
money however. Its simply a way to create two broadcast domains
where
there was previously one without additional replication of hardware
and
cabling.
"
You know, it seems that broadcasting is a lot like friction --
We spend a lot of time trying to reduce it, but we can't live without it
!
-------------------------------------------------
Tks | <mailto:[EMAIL PROTECTED]>
BV | <mailto:[EMAIL PROTECTED]>
Sr. Technical Consultant, SBM, A Gates/Arrow Co.
Vox 770-623-3430 11455 Lakefield Dr.
Fax 770-623-3429 Duluth, GA 30097-1511
=================================================
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
Bob Vance
Sent: Tuesday, January 16, 2001 12:50 PM
To: CISCO_GroupStudy List (E-mail)
Subject: RE: why is routing needed with VLANs - ARP?
What I'm saying is that, before we implement VLANs, we have a flat
address space, with obviously, no routing.
Now, suppose that I arbitrarily decide not to forward broadcasts out
ports 6-10 through some IOS command.
Everything will still work quite happily (except anything relying on
those broadcasts, of course).
...
Ooops. I think that I just saw the answer.
One of those broadcast thingys is lil' ole ARP.
So, how does a client find the IP address of a destination if the
destination is outside the VLAN?
It's funny that this wasn't pointed out in any of my VLAN reading
(admittedly limited to ICND coursebook and Caslow).
It just arbitrarily says unicasts are blocked or routing is
required without giving a reason.
Oh, well.
-------------------------------------------------
Tks | <mailto:[EMAIL PROTECTED]>
BV | <mailto:[EMAIL PROTECTED]>
Sr. Technical Consultant, SBM, A Gates/Arrow Co.
Vox 770-623-3430 11455 Lakefield Dr.
Fax 770-623-3429 Duluth, GA 30097-1511
=================================================
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
Bob Vance
Sent: Tuesday, January 16, 2001 11:35 AM
To: CISCO_GroupStudy List (E-mail)
Subject: why is routing needed with VLANs
OK.
I must be brain dead, today.
(and, yes, Chuck, I *have* had my morning dose of Diet Coke :)
and, yes, I know, "What's so special about 'today' "?
)
As far I can understand it so far, about the only benefit that I see
from VLANs is reducing the size of broadcast domains.
Suppose that I have a switch in the closet with one big flat address
space (well, it couldn't be that big with only one switch, now, could
it ?>). Then someone says,
"You know, we're getting a lot of blah-blah broadcast traffic.
Let's VLAN.
"
OK, fine. We VLAN and put whatever services in each VLAN that are
required to handle the broadcasts (e.g., DHCP service). So, now the
switch doesn't send broadcasts outside a particular VLAN.
But, what's so magic about a VLAN that the switch also decides not to
send unicasts outside a VLAN. Before the VLANs, the switch maintained
a MAC table and knew which port to go out to get to any unicast address
in the entire space. So, why can't it continue to do that after we
arbitrarily implement some constraint on broadcast addresses?
It seems to me that the same, exact MAC table, with an additional VLAN
field would not require that restriction. If it's a broadcast, send the
packet only out ports with a VLAN-id that matches the source port's
VLAN-id. If it's a unicast, handle it just like we used to.
Similarly, even if we have 5 switches, I just don't see the requirement
that we (as switch-code designers) must block unicasts and resort to a
routing requirement.
Even with 500 switches ... well, let's not get ridiculous :)
I feel that there is a simple point that I've overlooked, so I will
continue to RTFM while I await your responses.>)
-------------------------------------------------
Tks | <mailto:[EMAIL PROTECTED]>
BV | <mailto:[EMAIL PROTECTED]>
Sr. Technical Consultant, SBM, A Gates/Arrow Co.
Vox 770-623-3430 11455 Lakefield Dr.
Fax 770-623-3429 Duluth, GA 30097-1511
=================================================
_________________________________
FAQ, list archives, and subscription info:
http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
_________________________________
FAQ, list archives, and subscription info:
http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
