[RPSEC] Comments on draft-ietf-rpsec-bgpsecrec-07

[Joe Touch <[e-mail]>]

Hi, all,

Sorry for the belated reply to this draft; it was only recently brought
to the attention of the TCP-MD5 update design team in the TCPM working
group.

I hope these comments will be useful, and will track responses to this
post on the rpsec wg mailing list.

Joe

------------------------------------------

Abstract - Securing the info between the parties is not an
easy technical matter; if it were, IPsec would be in regular use. The
current situation is that TCP-MD5 (created for this purpose) has
weaknesses (RFC4808), and that there is IPsec is not used for this
(draft-ietf-tcpm-tcp-antispoof). The TCPM WG has created a design team
to explore two current proposals (draft-touch-tcpm-tcp-simple-auth,
draft-bonica-tcp-auth).

1.2 - "Attacks that do not involve..." is contrary to section 10 pg 15,
as well as the abstract itself.

10 - GTSM is not a security mechanism; it is an heuristic that happens
to work when tunneling is suitably secured (i.e., when there are
uncontrolled tunnels from a host, hopcount checking is useless). IPsec
should have been cited, certainly more prominently than GTSM.

It's confusing to propose that transport protection "MAY" be a
component; BGP without graceful restart treats transport failure as
routing failure (as noted in draft-tcpm-tcp-antispoof), and there are
substantial problems with graceful restart (as noted in RFC4724), both
of which should be summarized here.

-- 
----------------------------------------
Joe Touch
Sr. Network Engineer, USAF TSAT Space Segment

Hi, all,

Sorry for the belated reply to this draft; it was only recently brought
to the attention of the TCP-MD5 update design team in the TCPM working
group.

I hope these comments will be useful, and will track responses to this
post on the rpsec wg mailing list.

Joe

------------------------------------------

Abstract - Securing the info between the parties is not an
easy technical matter; if it were, IPsec would be in regular use. The
current situation is that TCP-MD5 (created for this purpose) has
weaknesses (RFC4808), and that there is IPsec is not used for this
(draft-ietf-tcpm-tcp-antispoof). The TCPM WG has created a design team
to explore two current proposals (draft-touch-tcpm-tcp-simple-auth,
draft-bonica-tcp-auth).

1.2 - "Attacks that do not involve..." is contrary to section 10 pg 15,
as well as the abstract itself.

10 - GTSM is not a security mechanism; it is an heuristic that happens
to work when tunneling is suitably secured (i.e., when there are
uncontrolled tunnels from a host, hopcount checking is useless). IPsec
should have been cited, certainly more prominently than GTSM.

It's confusing to propose that transport protection "MAY" be a
component; BGP without graceful restart treats transport failure as
routing failure (as noted in draft-tcpm-tcp-antispoof), and there are
substantial problems with graceful restart (as noted in RFC4724), both
of which should be summarized here.

-- 
----------------------------------------
Joe Touch
Sr. Network Engineer, USAF TSAT Space Segment

_______________________________________________
RPSEC mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/rpsec