RE: Reflecting new-MAM/SAM definition in diff-te drafts
"Francois Le Faucheur (flefauch)" <[email protected]>
| Newsgroups | gmane.ietf.tewg |
|---|---|
| Message-ID | <[email protected]> |
Hello Sandy, >> -----Original Message----- >> From: sanford goldfless [mailto:[email protected]] >> Sent: 20 May 2003 20:55 >> To: [email protected] >> Subject: RE: Reflecting new-MAM/SAM definition in diff-te drafts >> >> >> Francois, >> >> At this time, could you please clear up one point regarding >> overbooking >> with RDM? >> >> Max Reservable is now back as the choice for the parameter >> indicating >> the aggregate overbooking factor. Yes, but note that this was the asumption already in current RDM draft. >>You also show nicely that it can >> indeed be used in a CAC rule together with per-CT LOMs, by >> normalizing >> the per-CT reservations and then comparing to Max >> Reservable. A setting >> of Max Reservable could be used to enforce a lower aggregate >> reservation >> amount than the sum of the normalized per-TE-Class advertisements of >> Unreserved would. >> >> So, therefore, doesn't the semantic of this parameter and that of BCO >> now need to be kept distinct? I say this because it was my >> understanding that the advertised bw constraints would not >> be adjusted >> for link overbooking. Actually, again here, we need to distinguish the overbooking related to LSP/Link Size Overbooking and realted to the per-CT Link Overbooking Multipliers. Advertised BCs are not adjusted to reflect the Local Overbooking Multipliers, nor is Max Reservable Bw. But advertised BCs do inherently reflect the LSP/Link Size overbooking factors (as discussed in previous message to Jerry, BCs are "normalised" bandwidth which factor in the Link Size Overbooking, just like Max Reservable Bandwidth). Refering to section "6.3 Example Usage of LOM" of diff-te-russian-01: -the advertised BCs are BC0=200 and BC1=100 (regardless of the LOM values). - However, it may very well be that the actual link bandwidth is in reality only of 100 and that BC0=Max-Res-Bw effectively=200 already factors in an overbooking factor of 2. So I think BC0 does coincide with Max Reservable Bw (in the case of RDM only of course). Thanks Francois >>Or is this not right, and the BCs >> themselves in >> RDM must somehow be adjusted for overbooking? >> >> Thanks, >> Sandy Goldfless >> >> Sanford Goldfless >> 192 Fuller St >> Brookline MA 02446 >> 617-738-1754 >> [email protected] >> >> -----Original Message----- >> From: [email protected] [mailto:[email protected]] On >> Behalf Of Francois Le Faucheur (flefauch) >> Sent: Tuesday, May 20, 2003 12:32 PM >> To: Ash, Gerald R (Jerry), ALABS >> Cc: [email protected] >> Subject: RE: Reflecting new-MAM/SAM definition in diff-te drafts >> >> Jerry, >> >> >> >> >> Yes, we need a formula to make explicit the 'implicit' >> >> constraint (and for all the BC models, not just MAM). >> >> >> >> That's a good point. >> >> With RDM, the proposed formula of: >> >> o SUM (Normalized(CTc)) <= Max Reservable Bandwidth, >> >> for all "c" in the range 0 <= c <= (MaxCT-1) >> >> already applies, since we have: >> - SUM (Normalised(CTc)) <= BC0, and >> - BC0=Max Reservable Bw >> >> It would be nice to have exact same formula also apply to MAM. >> >> Cheers >> >> Francois >> >> >> >> >>