Re: [dc] Requirement for a method to manage mac address in DC

[David Allan I <[e-mail]>]

HI Fisher:

The size of the network is independent of the size of the VLAN. And I assume a larger network simply has more VLANs. So IMO the problem does not assert itself in proportion to network size.

MAC conflict is perhaps true when vendors manufacture many NICs with the same MAC and we have lots of experience with that. When the MAC is administered, and within a single administered broadcast domain,, a.k.a. VLAN I still assert problems will only arise via incompetence.

SuperVLAN looks to me like asymmetric VID or ETREE. Which is the one scenario when multiple MAC administrations MAY share a single subnet, although I suspect the cloud model where ETREE would  be used would preclude that (e.g. SaaS). If one is going to use ETREE for multiple IaaS instances, buyer beware...

cheers
Dave

________________________________
From: [e-mail] [mailto:[e-mail]]
Sent: Saturday, February 11, 2012 11:40 PM
To: David Allan I
Cc: [e-mail]; Lizhong Jin; [e-mail]
Subject: ´ð¸´: RE: [dc] Requirement for a method to manage mac address in DC


Hi David£º
   In an Large L2 network£¬the MAC conflict is likely£¬especially when mutil-vlan interwork through supervlan.
in addition£¬the benefit is not only avoiding the MAC conflict but also getting more controlling of their DC¡£

fisher.


>-----------------------------------------<
¨q¡Ð¨r¡è¡¡¡¡¡¡¡¡¡¡     Innovation change
                         the world
¨q¨q ¡Ð¨r        ¡ñ¨q¡ð¨r¡¡
¨t ----¨s       /¨€¡Å¨€\ ¡¡
~~~~~~~~~~~~~~~~~¡Ç~~¡Ç~~~~~~~~~~~~~~~~~
          My nickname: Fisher Yu
>----------------------------------------<

David Allan I <[e-mail]> дÓÚ 2012-02-12 06:27:38:

> Hi Lizhong
>
> VM MACs are administered, and only require uniqueness within a
> customer VLAN. So as long as a single customer VLAN does not have
> multiple administators, there should be no issue.
>
> IMO a MAC conflict within a single VLAN administration is complete
> incompetence, not somthing to kill ourselves trying to find a solution over!
>
> my 2 cents
> Dave
>
> From: [e-mail] [mailto:[e-mail]] On Behalf Of
> Lizhong Jin
> Sent: Thursday, February 02, 2012 6:47 AM
> To: [e-mail]
> Cc: [e-mail]; [e-mail]
> Subject: Re: [dc] Requirement for a method to manage mac address in DC

> Hi Thomas,
> The VMs connected by SPB-M or NVO3/VXLAN/NVGRE will still reside in
> one L2 network, and the VMs MAC will still conflict if have same MAC
> address. VM MAC confliction could not be mitigated or solved by
> these mechanism if I understand correctly.
>
> Lizhong
>
> -------------------------------------
> yu.jinghai at zte.com.cn writes:
>
>
> >
> >    I wonder whether it is necessary to manage mac address in DC.
> > As you know,VM's mac is randomly generated.
> > The risk of mac conflict is increasing with  the amount of VMs in DC.
> > If there is a method to auto manage and allocate mac address,the
> >    risk maybe avoid.
> > That method may facilitate the operator to control network easily
> >    and other available benefits.
>
> One straightforward way to mitigate C-MAC conflicts is via
> encapsulations, whether through something like SPB-M or
> NVO3/VXLAN/NVGRE.
>
> Do such approaches not address the problem sufficiently?
>
> Thomas