«

hi all

a customer wants to migrate a tokenring-lan (a private c-range net) onto
ethernet. On a separate eth-segment is the DHCP-server, wich acts as the
DNS-srv too.
The migrations should go as smoothly as possible. Due to lack of human
resources the migration will last several days (maybe weeks ???)

My idea was to connect a router - configured as SR/TLB - in between.
Therefor the migration can be done smootly.

unfortunately the DHCP and ARP request fail. I know.
Cause of the cannonical/non-canonical format

q: what is best practice to do such a migration without readressing the
whole subnet? and keeping all the current services up and running?
what are your experiences and suggestions?

tnx
stephan

stephan wrote:
> hi all
>
> a customer wants to migrate a tokenring-lan (a private c-range net) onto
> ethernet. On a separate eth-segment is the DHCP-server, wich acts as the
> DNS-srv too.
> The migrations should go as smoothly as possible. Due to lack of human
> resources the migration will last several days (maybe weeks ???)
>
> My idea was to connect a router - configured as SR/TLB - in between.
> Therefor the migration can be done smootly.
>
> unfortunately the DHCP and ARP request fail. I know.
> Cause of the cannonical/non-canonical format
>
> q: what is best practice to do such a migration without readressing the
> whole subnet? and keeping all the current services up and running?
> what are your experiences and suggestions?
>
> tnx
> stephan

check this
http://www.cisco.com/en/US/partner/t...8012811e.shtml
roland

"stephan" <> wrote in message
news:dortpr$pti$...
> hi all
>
> a customer wants to migrate a tokenring-lan (a private c-range net) onto
> ethernet. On a separate eth-segment is the DHCP-server, wich acts as the
> DNS-srv too.
> The migrations should go as smoothly as possible. Due to lack of human
> resources the migration will last several days (maybe weeks ???)

which is a real bad idea. better to wait until the people can do it and do a
"big bang" change.

if it is so important than you should be planning to test this. the test
resource for several migration steps probably dwarfs the amount of effort
needed for 1 big change.

otherwise you risk breaking things in wierd ways without understanding how
to fix them during migration.
>
> My idea was to connect a router - configured as SR/TLB - in between.
> Therefor the migration can be done smootly.
>
> unfortunately the DHCP and ARP request fail. I know.
> Cause of the cannonical/non-canonical format

from what i remember the various routers that supported translate bridge
only supported the protocols that they didnt handle via routing. to handle
translation bridging of IP you probably need a device that doesnt handle
routing.

If you want "real" translation bridging - find a proper translate bridge -
an old IBM one is probably the preferred way, since they were the standard
everyone else was measured against.

>
> q: what is best practice to do such a migration without readressing the
> whole subnet?

wrong assumption - it is best to readdress.

And if you have DNS or other logical naming for services then it should be
relatively straightforward.

> and keeping all the current services up and running?

test everything after every migration step - which implies only a few
(ideally only 1) step.

> what are your experiences and suggestions?

translation bridge is a black art, obsolete in most networks and no longer
widely used.

if you dont already know this stuff in detail - avoid it since learning by
trial and error is slow and expensive.
>
> tnx
> stephan
--
Regards

- replace xyz with ntl

stephen a écrit :
> "stephan" <> wrote in message
> news:dortpr$pti$...
>
>>hi all
>>
>>a customer wants to migrate a tokenring-lan (a private c-range net) onto
>>ethernet. On a separate eth-segment is the DHCP-server, wich acts as the
>>DNS-srv too.
>>The migrations should go as smoothly as possible. Due to lack of human
>>resources the migration will last several days (maybe weeks ???)
>
>
> which is a real bad idea. better to wait until the people can do it and do a
> "big bang" change.
>
> if it is so important than you should be planning to test this. the test
> resource for several migration steps probably dwarfs the amount of effort
> needed for 1 big change.
>
> otherwise you risk breaking things in wierd ways without understanding how
> to fix them during migration.
>
>>My idea was to connect a router - configured as SR/TLB - in between.
>>Therefor the migration can be done smootly.
>>
>>unfortunately the DHCP and ARP request fail. I know.
>>Cause of the cannonical/non-canonical format
>
>
> from what i remember the various routers that supported translate bridge
> only supported the protocols that they didnt handle via routing. to handle
> translation bridging of IP you probably need a device that doesnt handle
> routing.

Not true. Cisco handles bridging routable protocols. If you really
want to go that route, look at the "bridge irb" command. This will
allow to bridge IP across the two networks, but does not solve the DHCP
problem.

>>q: what is best practice to do such a migration without readressing the
>>whole subnet?
>
>
> wrong assumption - it is best to readdress.
>

Seconded.

if your /24 is less than half full, you can always split it in two,
change your DHCP scopes, wait for every one to obtain new addresses and
then assign the second half to the ethernet segment.

> And if you have DNS or other logical naming for services then it should be
> relatively straightforward.
>

Even then, you'll have people with hard coded mappings, printers, etc...

>
>>and keeping all the current services up and running?
>

Expect the unexpected and make sure that management is well aware that
the wheels may turn a little square for a while.


>
> test everything after every migration step - which implies only a few
> (ideally only 1) step.
>

Yup.

> if you dont already know this stuff in detail - avoid it since learning by
> trial and error is slow and expensive.

Yup again.

--
Francois Labreque | Unfortunately, there's no such thing as a snooze
flabreque | button on a cat who wants breakfast.
@ | - Unattributed quote from rec.humor.funny
videotron.ca