Hello Qiong, On 2011/07/28, at 2:28, Qiong wrote:
> Dear all, > > Following the recent discussion on stateless 4v6 solutions, I'm still > wondering how to define or plan a reasonable stateless 4v6 domain in reality. > My main concern is how to distinguish between multiple different domains with > lower cost ? > > I have noticed that we can define multiple stateless 4v6 domains in > draft-dec-stateless-4v6, 4rd, etc., within the same local-area. This can be > very helpful to apply different strategies for our customers. For example, > our legacy users can just have non-shared public IPv4 address while our new > customers would have shared-IPv4 address, or even with different multiplex > ratio. In this situation, multiple domains can be defined for different type > of users and need to be distinguished accordingly. > > Here, the ideal way is to define one IPv4 pool for one stateless 4v6 prefix. > However, our current situation is quite different. Not only does it has been > too difficult to retrieve the existing IPv4 addresses and re-plan them again, > but also the remaining IPv4 prefix pool is quite scattered. As a result, the > larger the domain covers, the more IPv4 prefix becomes. So, it might be > normal to have up to thousands of rules indicating specific IPv4 prefix pool > and corresponding IPv6 prefix in stateless gateway. Every packet arriving at > the stateless gateway has to firstly lookup the rule table and find its 4v6 > domain. This seems have turned a totally stateless solution into partial > stageful. Routing always supports us. You can choose a tunnel which have appropriate mapping rule by your routing table. Even it is 'totally stateless solution', you have to lookup routing table as far as it is routing network. Hope it helps your network design. Best regards, --satoru _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
