you are right. The algorithm needs to be different for the adj-sid. It is part 
of the changes I’m working at for the next revision of the draft.

Typically, the “action” in the algorithm (for the adj-sid) consists of a 
forwarding instruction out to the interface the adj-sid is allocated to.

s.


> On Feb 10, 2017, at 4:18 PM, Veerendranatha Reddy Vallem 
> <[email protected]> wrote:
> 
> Dear Previdi,
> SRH may carry Adj Segment and Special segments (128 IPv6 prefixes)  along 
> with Node Segments.
> 
> Ex: Consider below Topology
> 
> 
>                        / B ------ C------- D\                                 
>                                                                   
>                      /        Fadj| |                 \                       
>                                                                          
> X --------A                   | |                 H----- Y                    
>                                                                    
>                   \                   | |              /                      
>                                                                          
>                     \   E ----  F ----- G  /                                  
>                                                                  
> 
> 
> 
> 
>  At Node X                    At Node A,                                      
> At node C,     while forwarding to F (C need to decrement segments to twice, 
> one for Fadj and other for F)                                                 
>                                          
> IPv6 Hdr                              Ipv6 Hdr                                
>                  IPv6 Hdr
> DA= Y SA=X                      DA= C SA=X                                    
>         DA= F SA=X                                                            
>                               
>                                    SRHdr (segment left 4)                     
> SRHdr (segment left 2)                                                        
>                       
>                                   Y,H,F,Fadj,C                                
>                 Y,H,F,Fadj,C                                                  
>                                       
> 
> While processing this type of SRH, I feel existing algorithm may be required 
> to modify to add recursive look up to process Adj Segment/Special segment as 
> below. (Based on above example)
> 
> Existing algorithm: (As per section 4.3), 
> 
>   1.   IF DA = myself (segment endpoint)
>   2.      IF Segments Left > 0 THEN
>              decrement Segments Left
>              update DA with Segment List[Segments Left]
>   3.      ELSE continue IPv6 processing of the packet
>                End of processing.
>   4.   Forward the packet out
> 
> Proposed Modification:
> 
>    1.   IF DA = myself (segment endpoint)
>   2.      IF   Segments Left > 0  THEN
>              decrement Segments Left
>               do
>               IF Segment List[Segments Left]= Adj-Segment/Special Segment 
> originated by myself
>                     Trigger the action as per Segment (Ex: if it is Adj 
> Segment, identify interface to forward, if segment is for the service,  
> trigger for service)
>                     Decrement Segments Left
>               ELSE 
>                      Update DA with Segment List[Segments Left]
>                       break;
>             while (Segments Left > 0)    
> 
>   3.      ELSE continue IPv6 processing of the packet
>                End of processing.
>   4.   Forward the packet out
> 
> If SRH is having combination of Node and Adj SID then recursive look up to be 
> 2 times maximum at each IPv6 SR supported nodes 
> If Service Segments are added to SRH, then may be multiple recursive look 
> ups, since it may require to handle multiple services by same node.
> 
> Please check and let me know your opinion.
> 
> Thanks & Regards,
> Veerendranath
>       
> 
> 
> 
> 
> 
> -----Original Message-----
> From: Stefano Previdi (sprevidi) [mailto:[email protected]] 
> Sent: 10 February 2017 15:02
> To: Veerendranatha Reddy Vallem <[email protected]>
> Cc: [email protected]; 
> [email protected]; 
> [email protected]; [email protected]
> Subject: Re: [IPv6 SR] Regarding 128 bits IPv6 address in Segment List of SRH
> 
> Hi Veerendranath,
> 
> yes, an SR-IPv6 SID is a 128-bit IPv6 addresses. 
> 
> The semantic associated to the SID is given by the control plane. We have 
> already documented the signaling of the SIDs in ISIS, OSPF and BGP. Currently 
> we have defined Node-SIDs (representing a node) and Adjacency-SIDs 
> (instruction to forward out to the interface the SID is allocated to).
> 
> In the IPv6 dataplane a SID being an IPv6 address, it makes the SID a global 
> IPv6 address (even in the case of Adj-SIDs). This of course is orthogonal to 
> the control plane that may or may not advertise such address.
> 
> I.e., you may have an Adj-SID as a global IPv6 address that it is not 
> advertised by any routing protocol in the network (which implies of course 
> that the packet will have to first reach the node using a node-SID).
> 
> The use of LL addresses as SID has not been contemplated for the simple 
> reason that a router may well allocated the same address to all links so it 
> is not a reliable mechanism for forwarding. This will be fixed in the next 
> revision of the ospfv3 draft (isis draft is ok).
> 
> s.
> 
> 
> 
>> On Feb 10, 2017, at 7:37 AM, Veerendranatha Reddy Vallem 
>> <[email protected]> wrote:
>> 
>> Dear Previdi,
>> Thanks for your reply. 
>> 
>> As per first version of draft, Adj-SID is also  IPv6 prefix. 
>> 
>> 4.2.2.  Adjacency-SID
>> 
>>  The Adjacency-SID identifies a given interface.  In the SR
>>  architecture a node may advertise one or more Adj-SIDs allocated to a
>>  given interface so to force the forwarding of the packet (when
>>  received with that particular Adj-SID) into the interface, regardless
>>  the routing entry for the packet destination.  The same is defined
>>  for SR-IPv6: a node may advertise a given IPv6 prefix which is
>>  associated to the SR semantic of "send out the packet to the
>>  interface this prefix is allocated to".  Here also, the SID is in
>>  fact the IPv6 prefix.
>> 
>> As per my understanding  "segment list in SRH is global IPv6 prefixes, If we 
>> need to include Adj SID to segment list then Adj-SID is also global IPv6 
>> prefix". 
>> Please correct me if my understanding is wrong.
>> 
>> Thanks and Regards,
>> Veerendranath
>> 
>> 
>> 
>> 
>> -----Original Message-----
>> From: Stefano Previdi (sprevidi) [mailto:[email protected]] 
>> Sent: 09 February 2017 22:41
>> To: Veerendranatha Reddy Vallem <[email protected]>
>> Cc: [email protected]; 
>> [email protected]; 
>> [email protected]; [email protected]
>> Subject: Re: [IPv6 SR] Regarding 128 bits IPv6 address in Segment List of SRH
>> 
>> Hi,
>> 
>> the first version of the draft (draft-previdi-6man-segment-routing-header) 
>> had a description of Node-SID and Adj-SID. Later, in order to simplify the 
>> document, we removed the descriptions and focused the document into the SRH 
>> format.
>> 
>> I think it will be helpful to re-introduce a section on the two main SID 
>> types (Node, Adjacency). I’m working on an update for the next version.
>> 
>> Thanks.
>> s.
>> 
>> 
>> 
>> 
>> 
>>> On Feb 9, 2017, at 5:02 PM, Veerendranatha Reddy Vallem 
>>> <[email protected]> wrote:
>>> 
>>> Dear Authors,
>>> 
>>> I am requesting your clarification regarding usage of Adj-SID in SRH header
>> 
> 

_______________________________________________
OSPF mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ospf

Reply via email to