The MPLS-OPS Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Re: Why MPLS can't Guaranteed QoS?
It's not so much about MPLS or not it is the general issue with IP - no physical bandwith reservations - no intserv rsvp. All extra you can get with MPLS in this particular arena is to: A) engineer your entire network, B) allocate some room for the selected flows in the schduler. But once you have routing issues (and routing packets do get very high scheduling priority) and maybe even some loops your gurantees are the history, once you have the multicast flows - they can't be counted so again messing with your bandwith etc. It is true that with significat over-provisioning and carefull network design you can achive something what I call soft guarantees or in other words pretty good diffserv architecture. Hard guarantees are those which you see in ATM or in optical paths :). R. > Tarek Gad wrote: > > You mentioned in your response that MPLS can't guarantee QoS. May I > ask why? > > >From: fraanro > >To: raszuk@cisco.com > >CC: fraanro , Ruyter Hill , "'Dekany,Steven'" , > "'mpls-ops@mplsrc.com'" > >Subject: Re: Guaranteed QoS using MPLS? > >Date: Tue, 29 Jan 2002 08:26:52 GMT > > > >Well, I am sorry if I am wrong but, in my understanding of DS-TE, the > >difference with TE is that you may have more than one bandwitdh > pools, > >let us say one for each Diff serv class, but that does not mean any > per- > >LSP traffic scheduling (I have never heard that it is supported in > the > >current implementation, but probably I may be wrong), which IMHO is > >necessary to have guaranteed QoS. I think "soft" QoS with the right > >priorization for real time traffic (and limitation using for example > DS- > >TE) may be enough for almost all the applications over IP, but on the > >other hand, I would not say that MPLS can "guarantee" QoS today. > > > > > > > >Javier. > > > > > >----- Mensaje Original ----- > >Remitente: Robert Raszuk > >Fecha: Lunes, Enero 28, 2002 7:46 pm > >Asunto: Re: Guaranteed QoS using MPLS? > > > > > > > > I hope they do. > > > > > > I am also sure you know the difference between the control plane > > > reservation done for TE and actual scheduler's queue reservation > > > done in > > > DS-TE - right :) ? I had a feeling that some of this discussion > moves > > > around DS-TE as well. > > > > > > R. > > > > > > > fraanro wrote: > > > > > > > > > What you are > > > > > currently unable > > > > > to do with MPLS VPN which you can do with MPLS-TE is reserve > > > > > bandwidth > > > > > > > > Probably it is not necessary to say it, but for this > > > discussion...every> one is taking into account that currently, > > > MPLS-TE (at least the > > > > implementations of the main vendors like Cisco, Juniper, > > > > Riverstone, ...) does not actually "reserve" any bandwidth, it > is > > > > something that just happens in the control plane, not in the > > > forwarding> plane, right? > > > > > > > > Javier. > > > > > > > > ----- Mensaje Original ----- > > > > Remitente: Ruyter Hill > > > > Fecha: Lunes, Enero 28, 2002 4:01 pm > > > > Asunto: RE: Guaranteed QoS using MPLS? > > > > > > > > > Hi > > > > > > > > > > Ok I will clarify > > > > > > > > > > >you mentioned that MPLS TE tunnels are not very > > > > > >scalable. Would it be possible to be more specific and point > out > > > > > which>aspects of the TE tunnels are not scaling from an SP's > point > > > > > of view? > > > > > > > > > > I am not saying MPLS-TE is not scalable as such but that it > would > > > > > not be > > > > > scalable to have a separate TE tunnel for each of your classes > of > > > > > service > > > > > To have to constantly dimension the available bandwidth on > > > each of > > > > > thosetunnels, as traffic patterns change, would be a > management > > > > > headache to say > > > > > the least. > > > > > > > > > > >The second point I am asking is about the need for VPNs. The > > > original> > >question was about guaranteed QOS in MPLS - you seem > > > to be > > > > > suggesting that > > > > > >it is of limited value, > > > > > > > > > > I will clarify the point I was making. If you are to build an > > > MPLS-VPN > > > > > standard COS applies and in fact you can to a certain extent > map > > > > > DSCP AF and > > > > > DP into the EXP field of the MPLS shim header. What you are > > > > > currently unable > > > > > to do with MPLS VPN which you can do with MPLS-TE is reserve > > > > > bandwidth > > > > > > > > > > What I was saying would be nice to see is the ability to map > the > > > > > DSCP AF and > > > > > DP to the shim header on the MPLS VPN label then in turn map > > > that EXP > > > > > marking to a bandwidth guaranteed tunnel in my MPLS-TE core > > > > > > > > > > I hope this clarifies the issue > > > > > > > > > > Kind Regards > > > > > Hill > > > > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > From: Dekany, Steven [mailto:steven.dekany@marconi.com] > > > > > Sent: 28 January 2002 15:07 > > > > > To: Ruyter Hill; 'mpls-ops@mplsrc.com' > > > > > Subject: RE: Guaranteed QoS using MPLS? > > > > > > > > > > > > > > > HI Hill, > > > > > > > > > > > > > > > I read your comments with interest, since you obviously bring > > > into the > > > > > picture a service provider background and experience. I would > like > > > > > to follow > > > > > up on two points from your email. > > > > > > > > > > The first one is, that you mentioned that MPLS TE tunnels are > > > not very > > > > > scalable. Would it be possible to be more specific and point > > > out which > > > > > aspects of the TE tunnels are not scaling from an SP's point > of > > > > > view? Where > > > > > would you like to see changes for the better? Ease of > > > provisioning,> > hierarchy? > > > > > > > > > > The second point I am asking is about the need for VPNs. The > > > original> > question was about guaranteed QOS in MPLS - you seem > > > to be > > > > > suggesting that > > > > > it is of limited value, because it does not work with MPLS > VPNs. > > > > > Are you > > > > > suggesting that the majority of MPLS QOS services today > > > offered by > > > > > SPs, are > > > > > being offered via VPNs? > > > > > > > > > > Thanks in advance for any further thoughts, > > > > > > > > > > Best Regards, > > > > > > > > > > Steven > > > > > > > > > > -----Original Message----- > > > > > From: Ruyter Hill [mailto:Hill.Ruyter@carrier1.com] > > > > > Sent: Monday, January 28, 2002 8:17 AM > > > > > To: 'mpls-ops@mplsrc.com' > > > > > Subject: FW: Guaranteed QoS using MPLS? > > > > > > > > > > > > > > > > > > > > Hi > > > > > > > > > > I would like to comment here briefly > > > > > I had an interest in a similar functionality and found in > > > > > discussion with > > > > > Robert (thanks for the info Rob) > > > > > > > > > > That although you can guarantee bandwidth and pass policed > normal > > > > > IP traffic > > > > > on a MPLS-TE tunnel (not hugely scalable) > > > > > > > > > > You cannot at present pass already policed and marked packets > > > > > which are MPLS > > > > > labelled within an MPLS VPN over a specific bandwidth > guaranteed > > > > > tunnelbased on EXP field > > > > > > > > > > so if someone wanted to create an MPLS-VPN say for GRX > > > services and > > > > > guarantee SIP across it you would have to rely on normal > queuing > > > > > mechanismsand be sure to put enough fat in the network in > > > order to > > > > > guarantee bandwidth > > > > > is available > > > > > > > > > > Lets hope that soon we have the ability to do MPLS-VPN over > > > MPLS- > > > > > TE > > > > > > > > > > Regards > > > > > > > > > > Hill > > > > > > > > > > > > > > > -----Original Message----- > > > > > From: Christopher Lewis [mailto:chrlewis@cisco.com] > > > > > Sent: 28 January 2002 02:57 > > > > > To: saqibj@margallacomm.com > > > > > Cc: mpls-ops@mplsrc.com > > > > > Subject: Re: Guaranteed QoS using MPLS? > > > > > > > > > > > > > > > With the caveat that the amount of traffic the application > will > > > > > send is > > > > > known prior to the network being setup to service that level > of > > > > > traffic,yes. > > > > > > > > > > Try > > > http://www.cisco.com/warp/public/732/Tech/mpls/mpls_techdoc.shtml> > > > > > > > This link shows how this can be done on Cisco networks by > > > > > combining diff > > > > > serv QoS and MPLS traffic engineering capabilities. This > > > > > combination used > > > > > to be called Guaranteed bandwidth services, but was changed to > > > > > diff-serv > > > > > aware traffic engineering. For this to work properly, a > > > policer at > > > > > ingress > > > > > is necessary for the traffic eningeered tunnels to really > function > > > > > as you > > > > > want. > > > > > > > > > > Chris > > > > > > > > > > At 06:36 PM 1/27/2002, Saqib Jang wrote: > > > > > > > > > > > > > > > >Could MPLS be used to provide "virtual circuits" > > > > > >for IP applications having specific QoS requirements. > > > > > >For example, could MPLS be used to create guaratee QoS > > > > > >across an IP core for an application that requires > > > > > >no more that .1% packet loss? Do existing MPLS routers > > > > > >have such capabilities or would this require implementing > > > > > >a new MPLS standard? > > > > > > > > > > > >Also, how would an MPLS LER classify traffic that uses > > > > > >dynamic port numbers (e.g. SIP)? > > > > > > > > > > > >Saqib > > > > > > > > > > > >Margalla Communications, Inc. > > > > > >3301 El Camino Real, Suite 220 > > > > > >Atherton, CA 94027 > > > > > >(650) 298-8462 (W) > > > > > >(650) 274 8745 (C) > > > > > >(650) 368-8198 (F) > > > > > >saqibj@margallacomm.com > > > > > >http://www.margallacomm.com > > > > > > > > > > > >------- > > > > > >The MPLS-OPS Mailing List > > > > > >Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml > > > > > >Archive: http://www.mplsrc.com/mpls-ops_archive.shtml > > > > > > > > > > ------- > > > > > The MPLS-OPS Mailing List > > > > > Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml > > > > > Archive: http://www.mplsrc.com/mpls-ops_archive.shtml > > > > > > > > > > ------- > > > > > The MPLS-OPS Mailing List > > > > > Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml > > > > > Archive: http://www.mplsrc.com/mpls-ops_archive.shtml > > > > > > > > > > > > > > > This e-mail and any attachments are confidential. If you are > > > not the > > > > > intended recipient, please notify us immediately by reply > e-mail > > > > > and then > > > > > delete this message from your system. Do not copy this e-mail > > > or any > > > > > attachment, use the contents for any purposes, or disclose the > > > > > contents to > > > > > any other person: to do so could be a breach of confidence. > > > > > > > > > > ------- > > > > > The MPLS-OPS Mailing List > > > > > Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml > > > > > Archive: http://www.mplsrc.com/mpls-ops_archive.shtml > > > > > > > > > > ------- > > > > > The MPLS-OPS Mailing List > > > > > Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml > > > > > Archive: http://www.mplsrc.com/mpls-ops_archive.shtml > > > > > > > > > > > > > ------- > > > > The MPLS-OPS Mailing List > > > > Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml > > > > Archive: http://www.mplsrc.com/mpls-ops_archive.shtml > > > > > > >------- > >The MPLS-OPS Mailing List > >Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml > >Archive: http://www.mplsrc.com/mpls-ops_archive.shtml > > ---------------------------------------------------------------------- > Send and receive Hotmail on your mobile device: Click Here ------- The MPLS-OPS Mailing List Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml Archive: http://www.mplsrc.com/mpls-ops_archive.shtml
|
|