Multiprotocol Label Switching (mpls)

Muly Ilan | 17 Jun 2012 18:14

LSP ping as bootstrap for G-ACh encapsulated BFD

Hi,

We plan to implement the CC-CV-RDI functionality per RFC6428.

Is it mandatory to support LSP ping as a bootstrap for the BFD i.e. using LSP ping with TLV type 15 in the echo request/reply to exchange discriminator values?

RFC6428 is quite vague on this issue. It only states “Overall operation is as specified in RFC 5880 [4] and augmented for MPLS in RFC 5884 [8].”.

IMHO, the initial value of the remote discriminator can be zero and replaced with the correct value when the 1st control packet is received from the peer.

This behavior complies with another statement of the RFC “The transmitted Your Discriminator value MUST reflect back the received value of the My Discriminator field or be set to zero if that value is not known”.

Thanks,

Muly

_______________________________________________
mpls mailing list
mpls <at> ietf.org
https://www.ietf.org/mailman/listinfo/mpls

headers

Carlos Pignataro (cpignata | 4 Sep 2012 06:08

Re: LSP ping as bootstrap for G-ACh encapsulated BFD

What you describe seems to be the single-hop BFD initialization procedures, which are also used in RFC 5885 [1]

It is not totally clear to me how tightly tied the procedures from RFC 6428 are to the bootstrapping from RFC 5884 using LSP Ping, versus a single-hop initialization without discriminators exchanged. The document seems to be slightly vague there, though I do not know how it would affect.

Thanks,

-- Carlos.

[1] http://tools.ietf.org/html/rfc5885#section-3.1

o The BFD Control packets are sent on the VCCV control channel. The use of the VCCV control channel provides the context required to bind and bootstrap the BFD session, since discriminator values are not exchanged; the pseudowire demultiplexer field (e.g., MPLS PW Label or L2TPv3 Session ID) provides the context to demultiplex the first BFD Control packet, and thus single-hop BFD initialization procedures are followed (see Section 3 of [RFC5881] and Section 6 of [RFC5882]). o A single BFD session exists per pseudowire. Both PW endpoints take the Active role sending initial BFD Control packets with a Your Discriminator field of zero, and BFD Control packets received with a Your Discriminator field of zero are associated to the BFD session bound to the PW.

On Jun 17, 2012, at 12:14 PM, Muly Ilan wrote:

Hi,

We plan to implement the CC-CV-RDI functionality per RFC6428.

Is it mandatory to support LSP ping as a bootstrap for the BFD i.e. using LSP ping with TLV type 15 in the echo request/reply to exchange discriminator values?

RFC6428 is quite vague on this issue. It only states “Overall operation is as specified in RFC 5880 [4] and augmented for MPLS in RFC 5884 [8].”.

IMHO, the initial value of the remote discriminator can be zero and replaced with the correct value when the 1stcontrol packet is received from the peer.

This behavior complies with another statement of the RFC “The transmitted Your Discriminator value MUST reflect back the received value of the My Discriminator field or be set to zero if that value is not known”.

Thanks,

Muly

_______________________________________________
mpls mailing list
mpls <at> ietf.org
https://www.ietf.org/mailman/listinfo/mpls

_______________________________________________
mpls mailing list
mpls <at> ietf.org
https://www.ietf.org/mailman/listinfo/mpls

headers

Muly Ilan | 4 Sep 2012 11:24

Re: LSP ping as bootstrap for G-ACh encapsulated BFD

Thanks Carlos.

Yes, IMHO the procedure of RFC 5885 that you quoted below can be adopted also for MPLS-TP.

Assuming that in practice there is no PHP in MPLS-TP, the MPLS label provides the context to the first BFD control packet and there is no need for LSP Ping bootstrapping.

Muly

From: Carlos Pignataro (cpignata) [mailto:cpignata <at> cisco.com]
Sent: Tuesday, September 04, 2012 7:08 AM
To: Muly Ilan
Cc: mpls <at> ietf.org
Subject: Re: [mpls] LSP ping as bootstrap for G-ACh encapsulated BFD

What you describe seems to be the single-hop BFD initialization procedures, which are also used in RFC 5885 [1]

Thanks,

-- Carlos.

[1] http://tools.ietf.org/html/rfc5885#section-3.1

o The BFD Control packets are sent on the VCCV control channel. The

use of the VCCV control channel provides the context required to

bind and bootstrap the BFD session, since discriminator values are

not exchanged; the pseudowire demultiplexer field (e.g., MPLS PW

Label or L2TPv3 Session ID) provides the context to demultiplex

the first BFD Control packet, and thus single-hop BFD

initialization procedures are followed (see Section 3 of [RFC5881]

and Section 6 of [RFC5882]).

o A single BFD session exists per pseudowire. Both PW endpoints

take the Active role sending initial BFD Control packets with a

Your Discriminator field of zero, and BFD Control packets received

with a Your Discriminator field of zero are associated to the BFD

session bound to the PW.

On Jun 17, 2012, at 12:14 PM, Muly Ilan wrote:

Hi,

We plan to implement the CC-CV-RDI functionality per RFC6428.

Is it mandatory to support LSP ping as a bootstrap for the BFD i.e. using LSP ping with TLV type 15 in the echo request/reply to exchange discriminator values?

RFC6428 is quite vague on this issue. It only states “Overall operation is as specified in RFC 5880 [4] and augmented for MPLS in RFC 5884 [8].”.

IMHO, the initial value of the remote discriminator can be zero and replaced with the correct value when the 1stcontrol packet is received from the peer.

Thanks,

Muly

_______________________________________________
mpls mailing list
mpls <at> ietf.org
https://www.ietf.org/mailman/listinfo/mpls

_______________________________________________
mpls mailing list
mpls <at> ietf.org
https://www.ietf.org/mailman/listinfo/mpls