[sfc-dev] Service Chain Last Hop

Reinaldo Penno rapenno at gmail.com
Tue Aug 25 16:12:39 UTC 2015


You mean sending back to the original classifier? The decision should be as
decentralized as possible in order to scale. There are some situation and
protocols that do not lend themselves to such decentralization in the
current state of affairs but I consider that a opportunity to enhance SFC
instead of sending back to a central point.

From:  Ricardo Noriega De Soto <ricardo.noriega.de.soto at ericsson.com>
Organization:  Ericsson AB
Date:  Tuesday, August 25, 2015 at 6:02 AM
To:  Reinaldo Penno <rapenno at gmail.com>, "Zhou, Danny"
<danny.zhou at intel.com>
Cc:  "sfc-dev at lists.opendaylight.org" <sfc-dev at lists.opendaylight.org>
Subject:  Re: [sfc-dev] Service Chain Last Hop

    
 Is there any trick in the configuration to specify a SFF as the egress
point of the chain??
 
 I agree with Danny about the first option. I think it's a little bit more
elegant, otherwise, classifier should redirect traffic to the correct SFF
which would be in charge of encapsulating packets.
 
 What do you mean by Python SFF??? The SFC agent??
 
 The deeper I dig into this solution, the more confused I am... :)
 
 BR
 Ricky
 

 
 
 
On 08/25/2015 04:04 AM, Reinaldo Penno wrote:
 
 
>   
> We have implemented option 2 when SFC is used standalone. Each last SFF is
> responsible for sending plain packet after removing NSH. Some people want
> packet to go back to central point due to overlay or addressing issues.
>  
> 
>  
>  
> But you can ask Python SFF to send packet back to classifier (or anywhere
> really) by adding IP address to context header.
>  
> 
>  On Aug 24, 2015, at 6:57 PM, Zhou, Danny <danny.zhou at intel.com> wrote:
>  
>  
>  
>>  
>>    
>>  
>> 
>> I think you have two options:
>>  
>>  1)      Once the SFF detects the Service Index equal to zero, forward it to
>> classifier which then remove the NSH header and forward it to real
>> destination using inner headers.
>>  
>> 
>>  2)      SFF directly remove the NSH header and forward it to real
>> destination.
>>  
>> 
>>  
>>  
>> I personally prefer option No.1 as single point in the SFC to encap/decap NSH
>> headres.
>>  
>>  
>>  
>>  
>>  
>>  
>> 
>> From: sfc-dev-bounces at lists.opendaylight.org
>> [mailto:sfc-dev-bounces at lists.opendaylight.org] On Behalf Of Jo?o Silva
>>  Sent: Monday, August 24, 2015 11:24 PM
>>  To: sfc-dev at lists.opendaylight.org
>>  Subject: [sfc-dev] Service Chain Last Hop
>>  
>>  
>>  
>>  
>>  
>>  
>>  
>>  
>>  
>> 
>> Hi guys!  
>>  
>>  
>> In the current implementation what happens when the SFF receives the
>> vxlan+nsh packet from the last service function? I'm asking this because, it
>> seems the SFF is dropping the packet instead of delivering it to the its
>> destination.
>>  
>>  
>> Kind regards,
>>  
>>  
>> João 
>>  
>>  
>>  
>>  
>>  
>  
>>  
>> _______________________________________________
>>  sfc-dev mailing list
>>  sfc-dev at lists.opendaylight.org
>>  https://lists.opendaylight.org/mailman/listinfo/sfc-dev
>>  
>>  
>  
>   
>  
> _______________________________________________
> sfc-dev mailing list
> sfc-dev at lists.opendaylight.orghttps://lists.opendaylight.org/mailman/listinfo/
> sfc-dev
>  
 
 


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opendaylight.org/pipermail/sfc-dev/attachments/20150825/29f54d57/attachment.html>


More information about the sfc-dev mailing list