RE: Two questions on draft-ietf-ccamp-gmpls-ason-routing-ospf

[Igor Bryskin <[e-mail]>]

Dimitri,

But who said that linkID must be unique per routerID? Of course, not. It must be unique per TE Router ID. The fact that the same RC manages several TE RTRs does not change a thing. The paths are computed on the network TE graph, built of vertices (TE RTRs) interconnected by edges (TE links). TE links must be uniqely identifed either as numbered or unnumbered (combination of TERtrID and LinkID). How the links are advertised - by separate RCs, the same RC or via no RC (statically configured) - is of no importance.

Igor

[e-mail] wrote: igor

not an issue of TE router ID uniqueness, without this additional sub-TLV, 
an unnumbered local id may not be unique per router_id, hence the addition 
of this sub-TLV (TE Router ID being unique per router_id) 

-d





Igor Bryskin 
09/03/2007 22:36
 
        To:     Dimitri PAPADIMITRIOU/BE/ALCATEL@ALCATEL
        cc:     [e-mail], "Brungard, Deborah A, ALABS" 
, "Ong, Lyndon" , 
[e-mail]
        Subject:        RE: Two questions on 
draft-ietf-ccamp-gmpls-ason-routing-ospf


Dimitri,

I don't see how we can manage the situation where TENodeIds are not 
unique. At the end of the day we must process EROs/RROs that are build
of TENodeID/TELinkID pairs. If TELinkID has node-scope and TENodeID is not 
unique, how then we can identify the specified link? So what do you 
suggest we put in each of these two fields?

[e-mail] wrote:
igor 

but it doesn't solve the issue (and introduces different setting and 
processing of the link id value from rfc 3630) indeed, in ASON RC can be 
associated to multiple "nodes", each of these nodes can have overlapping 
id spaces (to identify the "links") 

for that reason the solution is that each TE link (top level) TLV has a 
new sub-TLV that associates the local and remote "node id" (the former and 

latter takes the TE Router ID as value)

it is the substance of what i have been pointing to you in my initial 
e-mail (see Section 5.2 but also in RFC 4652 Section 5.7: "the routing 
protocol MUST be able to disambiguate the advertised TE links so that they 

can be associated with the correct TE Router ID.")

-d.







Igor Bryskin 
Sent by: [e-mail]
09/03/2007 20:29

To: "Ong, Lyndon" , Dimitri 
PAPADIMITRIOU/BE/ALCATEL@ALCATEL
cc: [e-mail], "Brungard, Deborah A, ALABS" 
, [e-mail]
Subject: RE: Two questions on 
draft-ietf-ccamp-gmpls-ason-routing-ospf


Lyndon,

Let me try to explain my point. 

Basically, we have two solutions to address the situation when the 
relationship between a router controller and data plane switches (TE nodes 

or TE routers) are 1:N, that is, when a single controller manages multiple 

TE routers.

Solution 1 (suggested in the draft):

Each TE Link advertising is extended with Local/Remote TE Router IDs pair. 

In this case, what is in the LinkID sub-TLV is not important really.

Solution 2 (mine)

The Router Address TLV is extended to advertise all TE Router IDs 
controlled by the advertising controller as routable addresses. The TE 
Link TLV is extended to advertise the local TE Router ID. However, there 
is no need to advertise the remote TE Router ID, because this is the 
function of the existing LinkID sub-TLV, which is to identify the remote 
end of the advertised TE link – remote TE Router ID.

Both these approaches solve the problem; however, an important question to 

answer is:

Do we need the TE Router IDs to be routable addresses?

The answer is YES according to draft-ietf-ccamp-gmpls-addressing-05.txt. 
For example, if the network
is built of unnumbered TE links, than ERO signaled in RSVP-TE Path will 
contain
path in a form of TE_RtrID/ifIndex pairs, and having TE_RtrID routable 
solves the problem how to signal the Path message along the provisioned 
path..
If you agree with this, than you will also agree that all TE Router IDs 
should be advertised within Router Address TLV
Hope that this helps.

Igor




"Ong, Lyndon" wrote:
Hi Dimitri,

That was my understanding also, I don't see any issue with this.

Cheers,

Lyndon 

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf 

Of [e-mail]
Sent: Friday, March 09, 2007 7:56 AM
To: Igor Bryskin
Cc: [e-mail]; Brungard, Deborah A, ALABS; 
[e-mail]
Subject: Two questions on draft-ietf-ccamp-gmpls-ason-routing-ospf 

igor

the drafts says

"Note: The Link ID sub-TLV that identifies the other end of the link (i.e. 


Router ID of the neighbor for point-to-point links) MUST appear exactly 
once per Link TLV. This sub-TLV MUST be processed as defined in [RFC3630]. 


" 

which is exactly what you are saying - when i say "it identifies the 
remote RC not the remote data plane "node" in case the remote RC is 
associate to n nodes" read "it is set to the router_id that identifies the 

remote RC..."
in brief, we keep the semantic and add a discriminator (that does not 
apply in case of colocated 1:1 LSR) - this closes the first point -

concerning the second point, since there is a possibility to have multiple 

associations in different LSAs i don't where the problem is ?

thanks,
-d.





Igor Bryskin 
Sent by: [e-mail]
09/03/2007 16:05

To: Dimitri PAPADIMITRIOU/BE/ALCATEL@ALCATEL
cc: [e-mail], "Brungard, Deborah A, ALABS" 

Subject: Re: Two questions on 
draft-dimitri-ccamp-gmpls-ason-routing-ospf draft


Dimitri,

> Is the LinkID is the same as Remote TE Router ID? 

no

> LinkID unambiguosly identifies remote data plane node,

no, it identifies the remote RC not the remote data plane "node" in case 
the remote RC is associate to n "nodes"

IB>> No, I disagree. You see that's why it's so important to quote the
RFCs/drafts, because people often interpret them differently.

In RFC 3630 we read:
"
2.5.2. Link ID



The Link ID sub-TLV identifies the other end of the link. For

point-to-point links, this is the Router ID of the neighbor.

"

Note that it does not say whether this is the advertising Router ID 

(identifying neighbor RC) or TE Router ID (identifying the
neighbor TE node). However, it does say that it "identifies the other end 
of the link". Because a link is terminated by TE nodes (and not 
advertising routers) I conclude that LinkID identifies the remote TE node.

Furthermore, earlier in RFC 3630 we read:
"
2.4.1. Router Address TLV

The Router Address TLV specifies a stable IP address of the
advertising router that is always reachable if there is any
connectivity to it; this is typically implemented as a "loopback
address". The key attribute is that the address does not become
unusable if an
interface is down. In other protocols, this is known
as the "router ID"

I interpret that this is the same router ID as in the upper quote. It is 
advertised in the Router Address TLV, which is the only way today to 
advertise TE Router ID. Hence the router ID in the context of this RFC is 
the TE Router ID.

The conclusion #1: the TE Link TLV, as it is today, always contains the ID 


of the remote TE node. 

The conclusion #2: there is a need to advertise several TE Router IDs 
supported by the same RC (advertising router), which, I think, should be 
proposed in your draft

ps: second question is trivial, mathematical vs networking formulation (no 



real difference)

IB>> Well, it changes one of the fundamental definitions of G.8080, and I 
am asking why is that in the draft which is supposed to define ways to 
support G.8080 

Igor

pps: if you want to put guidelines on e-mail responses probably directing 
your e-mail to the GEN AREA would be more suitable 

hope this helps,
-d.




Igor Bryskin 
09/03/2007 00:03

To: Dimitri PAPADIMITRIOU/BE/ALCATEL@ALCATEL
cc: [e-mail], "Brungard, Deborah A, ALABS" 

Subject: Re: Two questions on 
draft-dimitri-ccamp-gmpls-ason-routing-ospf draft


Dimitri, no, it does not help.

You didn't answer the first question, which is:

Is the LinkID is the same as Remote TE Router ID? If no, what is the 
difference? If yes, why do you need the latter? Both your pointers explain 



why do you need advertising of the local TE Router ID (advertising router 
may cover multiple data plane nodes), However, LinkID unambiguosly 
identifies remote data plane node, and the need for the advertising of 
Remote TE Router ID is not obvious

BTW, You didn't answer the second question either.

Igor

PS, I have a suggestion for the working group: Let us disallow responding 
to the comments/questions by just pointing to RFCs or drafts. In my view 
it is quite unfriendly. It is always possible to cut and paste a piece 
from the relevant RFC or draft confirming the points the writer is trying 
to make.

[e-mail] wrote:
igor


pls use version (or 03 
when available to make comments)

in that version you will see in Section 5.2

" Note: The Link ID sub-TLV that identifies the other end of the link 
(i.e. Router ID of the neighbor for point-to-point links) MUST 
appear exactly once per Link TLV. This sub-TLV MUST be processed as 
defined in [RFC3630]. "

now why this sub-TLV 17, well for the reason explained at the beginning of 




par.5.2
but also in RFC 4652 Section 5.7

hope this helps,
-d.




Igor Bryskin 
08/03/2007 22:11

To: Dimitri PAPADIMITRIOU/BE/ALCATEL@ALCATEL
cc: [e-mail], "Brungard, Deborah A, ALABS" 

Subject: Two questions on 
draft-dimitri-ccamp-gmpls-ason-routing-ospf draft


Dimitri,
I have a couple questions wrt the 
draft-dimitri-ccamp-gmpls-ason-routing-ospf draft.
In 5.2 a TE Link sub-TLV is introduced for the purpose of advertising 
local and remote TE Router ID:

0 1 2 3 
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 
| 17 | Length | 
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 
| Local TE Router Identifier | 
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 
| Remote TE Router Identifier | 
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 

Although I do understand why there is a need for advertising the Local TE 
Router ID, I donââ'¬â"¢t understand why the Remote Te Router ID? 
Isnââ'¬â"¢t 
this is 
the same
information 
that is carried in the Link ID sub-TLV?
In 6. you write:

ââ'¬Å"A RA may contain smaller RAs inter-connected by links. 
The limit of the subdivision results in
a RA that contains two sub-networks interconnected by a single 
link.ââ'¬Â

In G.8080 I read:
ââ'¬Å".... A routing area is defined by a set of subnetworks, the 
SNPP 
links 

that interconnect them, and the SNPPs representing the ends of the SNPP 
links exiting that routing area. A routing area may contain smaller 
routing areas interconnected by SNPP links. The limit of subdivision 
results in a routing area that contains ]one subnetwork.ââ'¬Â

Why is the discrepancy?

Thanks,
Igor


[
Sucker-punch spam with award-winning protection.
Try the free Yahoo! Mail Beta.


Now that's room service! Choose from over 150,000 hotels 
in 45,000 destinations on Yahoo! Travel to find your fit.



Sucker-punch spam with award-winning protection.
Try the free Yahoo! Mail Beta.



Get your own web address.
Have a HUGE year through Yahoo! Small Business.


 8:00? 8:25? 8:40? Find a flick in no time
with theYahoo! Search movie showtime shortcut.



 
---------------------------------
Don't get soaked.  Take a quick peek at the forecast 
 with theYahoo! Search weather shortcut.