From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Fri Apr 30 2004 - 10:47:39 GMT-3
At 11:47 PM -0400 4/29/04, MMoniz wrote:
>Howard...when you have two physical eth int's on a router that connect to 2
>different switches
>but you want them on the same ip subnet how else can you do it?
>
>It is not backing up...it is converging. Say for instance I want 1 eth wired
>to this closet and the other eth to another closet. but all host on the same
>subnet. Would you tie the switches together? I wouldn't.
One of my general rules of network design is that as soon as you
start having to do weird and complex things to put things into the
same subnet, it's a message from the Universe saying "it's time to
split subnets and route." With modern equipment, the speed
difference between L2 and L3 just isn't significant, except
_possibly_ for two real-time switches.
What is the rationale for converging to the same address? If it's a
VLAN-capable switch, which most will be, make the two closets
separate VLANs that go to differently addressed subinterfaces on the
same physical interface?
>
>Set ALL host on both switches to the default gateway of the BVI int and they
>will communicate. This is the precise reason
>IRB was created. Arp will work as will bridging your normal netbios, etc.
>across the router.
>
>I don't understand why this is thought to be a backup solution.
I see it as useful for backup because, like HSRP with different
routers, it makes a failure transparent to the downstream devices.
>
>However if the originator of the question would specify more what he meant
>we can all come to a reasonable
>solution.
>
>But the original question stated.
>
>"I am trying to put the exact same IP Address on 2 different ethernet
>>>>>interfaces."
>
>I see no other feasable answer other than wanting to tie to seperate layer 2
>segments into 1 layer 3 segment.
>
>mike
>
>-----Original Message-----
>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
>Howard C. Berkowitz
>Sent: Thursday, April 29, 2004 11:16 PM
>To: ccielab@groupstudy.com
>Subject: RE: Same IP on E0/0 and E0/1
>
>
>At 7:16 PM -0400 4/29/04, MMoniz wrote:
>>that is why I said IRB or CRB is the only logical answer. Each int goes to
>>different lan switch but all host on both switches use the same default
>>gateway...which is the BVI int
>>
>>Only way that makes sense to me!!!
>>
>>could be wrong but that is my interpretation.
>>
>>So arp would work because of the BVI, just like in any LAN environment
>>having a "real" ip
>
>Good thinking -- that someone is using one router to back up two
>switches. While I've seen people try to do that, it needs LOTS of
>intelligence in the host to know which NIC to use (e.g., understand
>spanning tree), or the host has to send out duplicate traffic on both
>interfaces, in a manner that the application understands.
>
>I'm still confused about the problem, unless there is very
>substantial software development involved -- which probably makes the
>budget large enough to afford redundant routers.
>
>>
>>-----Original Message-----
>>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
>>Howard C. Berkowitz
>>Sent: Thursday, April 29, 2004 7:06 PM
>>To: ccielab@groupstudy.com
>>Subject: RE: Same IP on E0/0 and E0/1
>>
>>
>>At 6:22 PM -0400 4/29/04, MMoniz wrote:
>>>HSRP or VRRP do not put the same IP on the int's...they just use a
>>"virtual"
>>>address that the physical
>>>int assumes. You can still reach each int by their individual address..as
>I
>>>am sure you are aware of
>>>
>>>not the same thing as having each int have the same IP
>>
>>some things along these lines have been done in multiprocessor server
>>clusters. I'm still trying to understand why someone would want to
>>have the same IP address on two separate interfaces of the same
>>router, assuming that the goal is not L2 load sharing.
>>
>>It's often easier to make recommendations when the problem is known,
>>as opposed to how to make the router do something that seems very
>>undesirable from a protocol and router design perspective. Just as a
>>trivial example, how does ARP work if two interfaces have the same IP
> >address?
>>
>>>
>>>-----Original Message-----
>>>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
>>>Richard Dumoulin
>>>Sent: Thursday, April 29, 2004 6:05 PM
>>>To: Howard C. Berkowitz; ccielab@groupstudy.com
>>>Subject: RE: Same IP on E0/0 and E0/1
>>>
>>>
> >>For who not ?
>>>
>>>-----Original Message-----
>>>From: Howard C. Berkowitz [mailto:hcb@gettcomm.com]
>>>Sent: jueves, 29 de abril de 2004 23:45
>>>To: ccielab@groupstudy.com
>>>Subject: RE: Same IP on E0/0 and E0/1
>>>
>>>
>>>At 10:22 PM +0100 4/29/04, Richard Dumoulin wrote:
>>>>But are not hsrp and vrrp for backing up routers, not interfaces ?
>>>
>>>For Cisco, yes. And, if the problem is backup, it may be more
>>>important to think of different routers rather than different
>>>interfaces of the same router. The probability of one interface
>>>going down and the other staying up is, from my experience, less than
>>>that of losing the whole router.
>>>
>>>>
>>>>-----Original Message-----
>>>>From: Howard C. Berkowitz
>>>>[<mailto:hcb@gettcomm.com>mailto:hcb@gettcomm.com]
>>>>Sent: jueves, 29 de abril de 2004 23:14
>>>>To: ccielab@groupstudy.com
>>>>Subject: Re: Same IP on E0/0 and E0/1
>>>>
>>>>
>>>>At 12:47 PM -0700 4/29/04, RExpert wrote:
>>>>>I need help like right now - BIG Thanks to everyone that responds.
>>>>>
>>>>>I am trying to put the exact same IP Address on 2 different ethernet
>>>>>interfaces. The only thing that I could think of is - interface
>>>>>multilink 1, but it is not working, here is my config:
>>>>
>>>>What problem are you trying to solve by doing this? Some sort of
>>>>failover? Respond to a practice question?
>>>>
>>>>Is this something that HSRP or VRRP could not do?
>>>>
>>>> >
> >>>>interface Multilink1
>>>>> ip address 66.106.47.230 255.255.255.252
>>>>> ppp multilink
>>>>> multilink-group 1
>>>>>
>>>>>Ethernet 0/0
>>>>>
>>>>>no ip address
>>>>> ip load-sharing per-packet
>>>>> no ip mroute-cache
>>>>> ppp multilink
>>>>> multilink-group 1
>>>>>
>>>>>Ethernet 0/1
>>>>>
>>>>>no ip address
>>>>> ip load-sharing per-packet
>>>>> no ip mroute-cache
>>>>> ppp multilink
>> >>> multilink-group 1
This archive was generated by hypermail 2.1.4 : Mon May 03 2004 - 19:48:58 GMT-3