Dear Authors, Would like to remind on the following.
1) In section 3.3.2.1 (LAG/ECMP), the change suggested below articulates the inference of entropy in a much better way. >>The entropy information MAY be inferred from the NVO3 overlay header or >>underlay header. The entropy information should be ideally present in the overlay header e.g. Entropy Label for MPLS [RFC 6790], VXLAN – UDP source port. In case the overlay protocol does not support the necessary entropy information or the switches/routers in the underlay do not support parsing of the additional entropy information in the overlay header, the following support is needed in the underlay switches/routers. The underlay switches and routers should be programmable, i.e. select the appropriate fields in the underlay header for hash calculation based on the type of overlay header – the point to note is even the location of the underlay header in the packet will vary based on the type of overlay header. 2) In section 3.3.2.1 (LAG/ECMP), it is worth adding large flow load balancing as an optional requirement for optimal bandwidth utilization. http://datatracker.ietf.org/doc/draft-ietf-opsawg-large-flow-load-balancing/ Thanks, Ramki From: Jon Hudson [mailto:[email protected]] Sent: Wednesday, April 17, 2013 8:41 AM To: ramki Krishnan Cc: [email protected] Subject: Re: [nvo3] suggestions on data plane requirements draft +1 On Apr 17, 2013, at 5:25 PM, ramki Krishnan <[email protected]<mailto:[email protected]>> wrote: Dear Authors, In section 3.3.2.1 (LAG/ECMP), the change suggested below articulates the inference of entropy in a much better way. >>The entropy information MAY be inferred from the NVO3 overlay header or >>underlay header. The entropy information should be ideally present in the overlay header e.g. Entropy Label for MPLS [RFC 6790], VXLAN – UDP source port. In case the overlay protocol does not support the necessary entropy information or the switches/routers in the underlay do not support parsing of the additional entropy information in the overlay header, the following support is needed in the underlay switches/routers. The underlay switches and routers should be programmable, i.e. select the appropriate fields in the underlay header for hash calculation based on the type of overlay header – the point to note is even the location of the underlay header in the packet will vary based on the type of overlay header. Thanks, Ramki _______________________________________________ nvo3 mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/nvo3
_______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
