Cell Relay Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Re: [Q] Is this a queuing or traffic shaping problem?
I.371 says that OAM cells are aone of many possible causes for Cell Delay Variation (CDV) (others are Phy layer overhead, ATM layer muxing, AAL layer Muxing etc). UPC/NPC mechanisms should allow for the cell delay variation caused by OAM etc. This can be done by setting a large enough CDVT parameter for PCR. Thus, CDV introduced by the host and upstream networks is accounted for by the CDVT parameter in the UPC/NPC respectively. If the downstream networks have a reduced tolerance for CDV (I'm not sure why that would be the case, but I.371 takes it into consideration), then traffic shaping can be used to reduce the CDV of the outgoing stream (of course shaping would increase the delay). In article <33f8fa4d.48463667@news.demon.co.uk>, Andrew Valentine <andy@gargoyle.demon.co.uk> wrote: >Hi, > >If I had a switch which had numerous vc's entering it on a single port and >they all exited via a single port and I needed to inject OAM cells into all >these vc's simultaneously how would I do it so that I would not upset the >traffic contracts? > >Would I need to use some queueing scheme to ensure some form of fairness or >would traffic shaping on the individual channels be sufficient? > >Has anybody tried to develop a traffic shaper which can handle more than 1000 >different traffic contracts simultaneously? All the commercial ones seem to >be quite limited and can only handle around 50 different traffic contracts. > > >Many thanks > >Andy > -- -- |
|