Re: 答复: Questions about the draft of "A new Designated Forwarder Election for the EVPN"

["Satya Mohanty (satyamoh)" <[e-mail]>]

Haoweiguo,

In addition to Ali's clarifications.

Since EVPN is a BGP based solution, anytime the BGP peering between the PE and its corresponding peer (usually a Route Reflector), is down, the DF Election algorithm needs to be recomputed at all participating PEs. In fact the PEs could be connected through hierarchical RR clusters, and there can be a connectivity there as well.

These wont be visible to DRNI ; so it wont be able to act on these kind of failures.

Thanks,
--Satya



From: "Ali Sajassi (sajassi)" <[e-mail]<mailto:[e-mail]>>
Date: Tuesday, September 30, 2014 10:02 AM
To: Haoweiguo <[e-mail]<mailto:[e-mail]>>, "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: Re: 答复: Questions about the draft of "A new Designated Forwarder Election for the EVPN"


Haoweiguo,

First we are not in business of comparing/contrasting our solutions/procedures with the ones from other SDOs. We have written a requirement document [RFC7209], and we need to ensure our solutions to adhere to the specified requirements.

Second, wrt your question on what are the advantages of EVPN redundancy mechanism compare to DRNI, just FYI, there are couple of main advantages:

  1.  EVPN can accommodate any number of systems in a redundancy group (not just 2 or 3 !!) - e.g., in DRNI terminology, you can have any number of systems in a Portal
  2.  EVPN doesn’t require any data links (virtual or physical) between the member of redundancy groups – e.g., no Intra-Portal Links are required

Cheers,
Ali


From: Haoweiguo <[e-mail]<mailto:[e-mail]>>
Date: Monday, September 29, 2014 at 10:46 PM
To: Cisco Employee <[e-mail]<mailto:[e-mail]>>, "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: 答复: Questions about the draft of "A new Designated Forwarder Election for the EVPN"


Hi Ali,

Compared to IEEE DRNI, what's the advantage of your solution? It's better to add a section in your draft  to describe the comparison.

Thanks

weiguo

________________________________
发件人: Ali Sajassi (sajassi) [[e-mail]<mailto:[e-mail]>]
发送时间: 2014年9月30日 12:36
收件人: Haoweiguo; [e-mail]<mailto:[e-mail]>
主题: Re: Questions about the draft of "A new Designated Forwarder Election for the EVPN"


There is no plan to use DRNI for EVPN.

-Ali

From: Haoweiguo <[e-mail]<mailto:[e-mail]>>
Date: Monday, September 29, 2014 at 8:03 PM
To: "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: Questions about the draft of "A new Designated Forwarder Election for the EVPN"


Sorry to repost the email to make it more clear...



Hi Co-authors,
In all-active connection case, to avoid frame duplication and looping, only one exit point among multiple PEs corresponding to a ESI is allowed to egress BUM traffic from EVPN network to local access network . DF election algorithm is a candidated solution. To overcome some DF election drawbacks,this draft proposed a new algorithm to replace DF election algorithm.
IEEE802.1AX,DRNI specs also can meet EVPN active-active connection requirements. Do you have plan to use IEEE existing standard to solve the problem?
Thanks
weiguo
________________________________
发件人: L2vpn [[e-mail]<mailto:[e-mail]>] 代表 Haoweiguo [[e-mail]<mailto:[e-mail]>]
发送时间: 2014年9月30日 10:37
收件人: [e-mail]<mailto:[e-mail]>
主题: Questions about the draft of "A new Designated Forwarder Election for the EVPN"


Hi Co-authors,
In all-active connection case, to avoid frame duplication and looping, only one exit point multiple PEs corresponding to a ESI is allowed to egress BUM traffic from EVPN network to local access network among . DF election algorithm is a candidated solution. To overcome some DF election drawbacks,this draft proposed a new algorithm to replace DF election algorithm.

IEEE802.1AX,DRNI specs also can meet EVPN active-active connection requirements. Do you have plan to use IEEE existing standard to solve the problem?

Thanks

weiguo