Xuxiaohu,

Comment inline

Yours irrespectively,

John


> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On Behalf Of
> Xuxiaohu
> Sent: Thursday, September 06, 2012 6:27 PM
> To: Ivan Pepelnjak; 'Balus, Florin Stelian (Florin)'
> Cc: [email protected]; 'Somesh Gupta'
> Subject: Re: [nvo3] Support for multi-homed NVEs
> 
> If the DCVPN solution has a control plane for VPN membership auto-
> discovery, the concern of whether the destination endpoint (i.e.,
> egress PE or ETR) is reachable can be addressed easily. Of course, if
> you need fast convergence, you can use the mechanism such as next-hop
> tracking, just as what Yakov has mentioned before. In addition, in the
> case where the egress PE is still reachable but its connectivity to the
> multi-homed CE is lost, if there is a control plane for MAC withdrawal
> or IP route withdrawal, there would not be any black-holing issue.


JD:  Block MAC withdrawal in E-VPN occurs in exactly this case.  The egress PE 
uses the withdrawal of a single route to indicate that all MAC addresses 
associated with the multi-homed CE are no longer reachable via it.  This route 
is the Per ESI Ethernet AD route which represents the connectivity between the 
egress PE and the multi-homed CE and it is associated with all VPNs of which 
the multi-homed CE is a member.


> Anyway, any DCVPN solution which lacks such a control plane seems not a
> good choice, especially in the multi-homing scenario.
> 
> Best regards,
> Xiaohu
> 
_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to