Re: M3UA AS override mode from SGP perspective
Benjamin Wilson <[email protected]>
| Newsgroups | gmane.ietf.sigtran |
|---|---|
| Message-ID | <[email protected]> |
Guys, Just thought I'd get some further clarification on Mikhail's issue. I'm still a bit uncertain about what is expected at the SG. For example, how would the following be fully meshed sencenario be handled? ASP1 has associations to both SGP1 and SGP2. ASP1 sends ASPUP (using ASP id 1) to both SGPs. Both SGPs respond accordingly with ASPUP ACKs. ASP1 sends an ASPAC (Override mode) to SGP1. Are both SGPs now supposed to consider the ASP active at the AS through "intercommunication"? If ASP1 then sends an ASPAC to SGP2, does he override himself and cause a NOTIFY(ALT-ASP-ACTIVE) to be sent from SGP1? Or is the SGP supposed to know that this the same ASP and DATA messages should now be sent from BOTH SGPs to ASP1? Cheers, Ben From: Mikhail Plotkin <[email protected]> To: "Brian F. G. Bidulock" <[email protected]>, Barry Nagelberg <[email protected]>, <[email protected]> Date: 10/01/2010 04:17 PM Subject: Re: [Sigtran] M3UA AS override mode from SGP perspective Sent by: [email protected] Hi, The issue is clear now. Thank you for the explanations. BR/ Mikhail BFGB> Mikhail, BFGB> Yes, Barry has explained it. We decided long ago that SGP must BFGB> coordinate amoung themselves within an SG (it makes things much BFGB> easier at the ASP). That assumes some sort of communication if they BFGB> are otherwise independent. When they are at a significant BFGB> geographical distance, for redundancy, they would be better BFGB> implemented in separate SG. BFGB> --brian BFGB> Barry Nagelberg wrote: (Fri, 01 Oct 2010 08:58:12) >> Mikhail, >> >> The scenario that you have described is not valid - if SGP_1 and SGP_2 >> are both part of SG, then there must be inter-communication between >> them. Otherwise they are just two independent SGs. >> >> In override mode, from the viewpoint of the SG there will still be >> only one active ASP in the AS, regardless of whether or not SGP1 and >> SGP2 distributed across two hosts. >> Just thought I'd get some further clarification on Mikhail's issue. I'm still a bit uncertain about what is expected at the SG. For example, how would the following be fully meshed sencenario be handled? ASP1 has associations to both SGP1 and SGP2. ASP1 sends ASPUP (using ASP id 1) to both SGPs. Both SGPs respond accordingly with ASPUP ACKs. ASP1 sends an ASPAC (Override mode) to SGP1. Are both SGPs now supposed to consider the ASP active at the AS through "intercommunication"? If ASP1 then sends an ASPAC to SGP2, does he override himself and cause a NOTIFY(ALT-ASP-ACTIVE) to be sent from SGP1? Or is the SGP supposed to know that this the same ASP and DATA messages should now be sent from BOTH SGPs to ASP1? >> Barry Nagelberg >> Adax, Inc. _______________________________________________ Sigtran mailing list [email protected] https://www.ietf.org/mailman/listinfo/sigtran _______________________________________________ Sigtran mailing list [email protected] https://www.ietf.org/mailman/listinfo/sigtran