Re: [mpls] Questions on BFD MPLS MIB

Hi,
Please find below the comments inline with the tag [VM].

Thanks,
Venkat.

On Sun, Jun 3, 2012 at 12:12 AM, Saravanan Narasimhan
<saravanannarasimhan@gmail.com> wrote:
> Hi Authors,
>
> I have a question regarding the Section 5.2.1 of the draft
> draft-vkst-bfd-mpls-mib-02.
>
> It states that
>
> " bfdMplsSessMapType = teIpv4(3),
>
>  -- OID of the first accessible object (mplsTunnelName) of
>  -- the mplsTunnelEntry identifying the MPLS TE tunnel (being -- monitored
> using BFD) in the MPLS tunnel table.
>  -- A value of zeroDotzero indicates that no association -- has been made as
> yet between the BFD session and the path
>  -- being monitored.
>  -- In the above OID example:
>  -- 100 -> Tunnel Index
>  -- 1 -> Tunnel instance
>  -- 3221225985 -> Ingress LSR Id 192.0.2.1
>  -- 3221225987 -> Egress LSR Id 192.0.2.3
>
>     bfdMplsSessMapPointer = mplsTunnelName.100.1.3221225985.3221225987,
>     bfdSessRowStatus = createAndGo
> }
>
> Similarly BFD session would be configured on the tail-end of the tunnel.
>
> "
>
> Here, at the tail-end of the tunnel, "Instance" is never really known.
> Tunnel can be setup with any tunnel instance (LSP ID) when it is signaled
> through RSVP-TE / CR-LDP. It need not be "1" as mentioned in the above
> example.
>
> So, We should be setting the map pointer as
>
> "bfdMplsSessMapPointer = mplsTunnelName.100.0.3221225985.3221225987". Am I
> right?

[VM] If I'm not wrong, tunnel instance 0 is reserved for configured
tunnel, we should be able to get the LSP-id through RSVP-TE or
out-of-band signalling (LSP Ping).

> Or should we take the approach as defined in PW-MPLS-STD-MIB (RFC 5602)
> where OutboundTunnel is mapped to the PW through individual objects
> (pwMplsOutboundTunnelIndex, pwMplsOutboundTunnelInstance,
> pwMplsOutboundTunnelLclLSR and pwMplsOutboundTunnelPeerLSR with
> pwMplsOutboundTunnelInstance as a READ-ONLY parameter) instead of a
> RowPointer approach done here. Please clarify.

[VM] RowPointer approach should be fine.

> Also, Can you please clarify how this MIB can be used to monitor a tunnel
> setup through 1:1 protection support (RSVP-TE signaling as in RFC 4872) (2
> LSP's -> Working and Protection)?
>
> Should we create 2 independent BFD sessions to monitor working and
> protection LSP or a single BFD session to monitor both?

[VM] Yes, two independent BFD sessions are required to monitor both
working and protection paths.

> If we need to create 2 BFD sessions, how to map tunnel pointer to BFD
> session?

[VM] Either you can configure tunnel pointers at the head and tail
nodes manually or use out-of-band signalling using on-demand CV.
>
> Please clarify.
>
> Thank You,
> N.Saravanan
>
> _______________________________________________
> mpls mailing list
> mpls@ietf.org
> https://www.ietf.org/mailman/listinfo/mpls
>