The MPLS WG Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Reg. MPLS-RSVP LSP Tunnels...
Hi Jie, If I understand Nallur's problem statement correctly, its got to do with aggregation of RSVP micro flows into LSP Tunnels setup by MPLS-RSVP. I guess, Routers A and B are RSVP(2205 only) aware and S and R are the end-hosts of a microflow. Hence there's no chance of receiving LROs in PATH messages generated from S. Aggregation of millions of RSVP micro-flows has been mooted in the Applicability Statement draft for RSVP LSP Tunnels. However, I too am not sure of an answer to this problem. Stating it in different words, the problem statement goes like this: An RSVP-LSP-Tunnel has already been established between LER1 and LER2. Now a micro-flow RSVP PATH message from S arrives at LER1. We want this micro-flow to take the LSP. So if we just tunnel this PATH message into the LSP, and make it reach end-host R, the problem is how do we transport the RESV message from R back to S through the same LSP since this is uni-directional? I hope Nallur agrees with this re-statement! All comments are welcome. -Jah -----Original Message----- From: Jie Zou <jzou@mars.iol.unh.edu> To: Pramoda Nallur <Pramoda.Nallur@ind.alcatel.com> Cc: mpls@UU.NET <mpls@UU.NET>; Dilip.Pandit@ind.alcatel.com <Dilip.Pandit@ind.alcatel.com>; Kishore Rao <kishore>; Manu.Prakash@ind.alcatel.com <Manu.Prakash@ind.alcatel.com> Date: Tuesday, July 25, 2000 6:49 PM Subject: Re: Reg. MPLS-RSVP LSP Tunnels... > >I hope the following can answer a part of questions. > >According the darft, if A or B is a non-RSVP router, it can't convey >labels via RSVP path. And it's neighber who knows it's non-RSVP >router should send a PathErr back to the sender. > >If A or B is an RSVP router but doesn't recognize the LRO, it should >send a PathErr toward the sender. > >If A or B is an RSVP router and recognize the LRO but not C_TYPE also >sends a PathErr toward the sender. > >Else the Path msg should pass A and B toward the receiver and >vice verse the Resv msg should pass the B and A toward the sender. > > >Regards, > >----------------------------------------------------- >Jie Zou (603)862-4212(Office) >MPLS Consortium >IOL, UNH (603)868-2983(Home) >----------------------------------------------------- > >On Tue, 25 Jul 2000, Pramoda Nallur wrote: > >> Hi, >> >> I have a query on setting up RSVP MPLS LSPs in an ATM network. >> [ Topology Driven LSP Setup Scenario ] >> >> Suppose the network configuration is as follows : >> >> S--A--LER1---LSR---LER2--B--R >> >> where, >> S --> Sender(RSVP aware) >> R --> Receiver(RSVP aware) >> A, B --> Legacy IP Routers >> LER1, LER2 and LSR --> MPLS-RSVP ATM switches. >> >> An LSP setup is initiated from LER1 with FEC=(S,R). >> >> The RSVP-TE PATH msg (with Sender=S, Session = Sess_R) creates an >> RSVP Session, Sess_R at LER1, LSR and LER2. The Egress LER2 responds with a >> >> RSVP-TE RESV msg reserving resources and sets up an LSP via LSR to LER1. >> >> After the LSP is setup, the RSVP-TE PATH msgs from LER1 and RSVP-TE RESV >> msgs >> from LER2 periodically refresh the RSVP reservation state on all nodes >> along the LSP. >> >> Suppose, Sender S sends an RSVP PATH msg to R (with Session = Sess_R). >> >> 1. Will this PATH msg take the unlabelled path OR will it get switched >> onto the LSP at LER1 ? >> >> 2. Also, how are the resulting RESV msgs from R propagated back to S ? >> >> 3. If these PATH and RESV msgs take the unlabelled path, will this result >> in >> resource allocation again along all nodes traversed by the LSP ? >> >> >> The "RSVP-TE: Extensions to RSVP for LSP Tunnels" >> (draft-ietf-mpls-rsvp-lsp-tunnel-06.txt) >> does not mention how the above should be handled. >> >> Any comments on this ? >> >> - Pramoda Nallur >> > > > ____________________________________________________________________ Get your own FREE, personal Netscape WebMail account today at http://webmail.netscape.com.
|
|