Re: [BEHAVE] Whether or not need to distinguish the primary and the backup NAT boxes?

[xuxiaohu 41208 <[e-mail]>]

Hi Shin,

----- 原邮件 -----
发件人: Shin Miyakawa <[e-mail]>
日期: 星期五, 三月 26日, 2010 上午1:40
主题: Re: [BEHAVE] Whether or not need to distinguish the primary and the backup NAT boxes?
收件人: [e-mail]
抄送: [e-mail]

> Xuxiaohu,
> 
> > Yes, the administrators can use them in that way. However, IMHO, 
> it is still better to assure the traffic of a given session 
> traverse the identical NAT box in normal case.
> 
> Usually one ISP network is big enough to have mutiple EBGP routers 
> to have
> redundancy, then, packets go through a box A and their responses
> return through the box B other than A.

Symmetric path usage is beneficial in some cases, e.g., traffic monitoring.

> So that's the reason why I said at the mic that it depends on the 
> network design.
> 
> Also, I think that "Cold Stand-by" means that the stand-by box is 
> just 
> "sleeping" usually. Even if you think "normal case" is for small 

No, the standby box is not sleeping, you can deploy redundancy and load-balancing together, just like what VRRP does for multiple VRRP groups. For more details, please see the load-balancing section of draft-xu-behave-stateful-nat-standby.

BR,
Xiaohu

> size of 
> Enterprise or so, it sounds like waste of investiment.
> We should be noticed that "running parallelly" often means "load 
> balancing" too.


 
> Best wishes,
> 
> Shin Miyakawa
>