Cell Relay Archive

[Martin Winter <Nospam.Martin.Winter@oen.siemens.de>]

Srikrishna wrote:

> Here I go again with PNNI. Please bear with me while I am exposing my
> total ignorance. It is not easy reading the PNNI spec. Hence the example
> below.

> I have a physical switch which is made up of two logical nodes. One of
> them is the "lowest level node" and the other is the "Logical group
> node". The functions of both these node are implemented in the same
> physical switch.

> Let us say,

> The "lowest level node" part is identified as AESA address A.2.x.y where
> A is 12 bytes, 2 takes up 1 byte, x is 6 bytes (ESI of the physical
> switch which implements the lowest level node function)  and y = 0 is 1
> byte (SEL) for a total of 20 bytes.

> The "logical group node" part is identified as AESA address A.2.x.1
> where A again is 12 bytes, 2 takes up 1 bytes, x is the same 6 bytes
> (ESI of the physical switch whihc implements the LGN function) and y = 1
> is 1 byte (SEL) for a total of 20 bytes.

> Before I continue, Is the above theory correct or am I already screwed
> up?

You're correct. That is a possible configuration.

> Anyway, assuming I am correct,

> Peer group ID of the lowest level node = 104.A.2 where 104 is the level
> of the peer group of which this "lowest level node" is a member of.

> Node ID of the lowest level node part = 104.160.A.2.x.0 for a a total of
> 22 bytes

> where 104 is the level of the "lowest level node's" containing peer
> group which is 13 bytes (made up of A.2), 160 indicates that this is the
> "lowest level node" and the remaining 20 bytes is the AESA address of
> the "lowest level node" as indicated before.

> Peer group ID of the LGN = 96.A where 96 is the level of the peer group
> of which this LGN is a member of.

> Node ID of the LGN = 96.104.A.2.x.0 for a total of 22 bytes
> where 96 is the level of the LGN containing peer group which is 12 bytes
> (made up of A), 104 indicates that it is a LGN of the peer group whose
> level is 104 and the remaining 20 bytesis the AESA address of the
> "lowest level node" as indicated before.

> As one can see that the AESA addresss are different for both the "lowest
> level node" and the LGN parts even though they are in the same physical
> switch. This difference is in the SEL value (o for lowest level node and
> 1 for the LGN) as dictated in section 3.3.1 on page 19.

> The LGN needs a AESA address obviously to setup an RCC SVC between which
> may or may not be directly connected. Why does the "lowest level node"
> need its own unique AESA address?. Is it because it may need to setup
> SPVC/SPVP other than that I do not see the need from a signalling point
> of view

You are right. There is no explicit need to assign an AESA address to
the lowest level node. PNNI will never setup a call using this AESA
address.

> Also the peer group ID's and the Node ID's are different as required. Am
> I correct that when the AESA address and the level information are
> configured into the switch, the peer group ID's and the node ID's are
> automatically generated.

No, not necessarily. The procedure described in the PNNI spec is only a
recommendation and not mandatory. A PNNI protocol stack implementation
could do it this way, but it would be more flexible, if all these IDs
could be configured separately.

> Can you confirm if I have derived these #'s above correctly given the
> initial AESA addresses.

Yes, you did it.
 
> How I am doing so far?. Now my main Q.
> 
> Even after talking out the above example out loud, I am not sure the
> real need for a node ID.

If the procedure described above would be mandatory, then you are right.
The node Id could be derived from the AESA address. But because it is
only a recommendation...

You could ask, why the ATM Forum did not make this mandatory. I do not
know the answer. But perhaps they were not sure about the consequences
for network configuration, if this would be mandatory. Consider an
existing network, where the nodes already have addresses which do not
fully conform to these recommendations. Would it still be possible to
create valid PNNI node Ids and PG Ids?
 
> Thanks for all the help in advance..
> 
> Srikrishna..
> 
> PS : The above represent my personal views NOT that of my employer.

-- 
Martin Winter

--------------------------------------------------------------------
 Siemens AG            Voice: +49 89 722-63718
 OeN ME 32               Fax: +49 89 722-26877
 Hofmannstr. 51
 81359 Muenchen         Mail:
 Germany            Internet: Martin.Winter@oen.siemens.de
--------------------------------------------------------------------