Re: Please Comment on New P2P VPN Problem Statement

Vishwas,

Thanks for your comments. I’ll respond inline below.

1. I guess the problem statement is not just about lessening the number of configuration commands but also the fact that static configuration may not work in some cases. The spokes may get new addresses every time they come up (using DHCP/ PPPoE) and hence the communication end point identifiers change.

SH> Yes, the dynamic nature of configuration data is as much a problem as the size of that data.

2. I am not sure but the use cases do not come out very clearly to me. The most important part of the communication is of end-sites communicating to the gateway hub router. In a typical enterprise deployment that would mean a branches connected to the campus/ data center. This tunnel is permanent. Mainly to access resources at the back end. There could be redundancy here to provide HA.

3. We then optionally require communication between end sites and such communication may be temporary or permanent. For such cases we want to be able to unburden the gateway so as to not cause overload.

4. We could have multiple gateways work in a cluster mode to serve a set of end-sites and to provide HA.

5. The clusters may in turn communicate with each other.

SH> I’m sorry that the use cases are not clear. However, I’m sure we can and will improve them.

SH> Your description of “The most important part of the communication” seems to be focused only on use case 2.2 in the draft. In your example, several “end-sites” need to establish a direct connection to each other instead of going through a “gateway hub router”. To use the terminology defined in the draft, that’s a gateway-to-gateway use case. I understand that you consider that to be the most important use case but others have previously spoken out about the importance of use cases 2.1 and 2.3.

SH> So I do agree with your description of use case 2.2. Maybe we should add another example under use case 2.2, based on your example. Could you provide more details about why a direct connection between two “end-sites” might be needed? I can add that as an example in the next version of the draft.

SH> And thanks for volunteering to participate in formulating the problem statement and the solutions. That’s great!

Take care,

Steve