Re: [tcpm] WGLC: draft-ietf-tcpm-tcpsecure-10.txt

["Anantha Ramaiah (ananth)" <[e-mail]>]

Ted,
  Appreciate the detailed comments. Pl see inline responses.


> -----Original Message-----
> From: Ted Faber [mailto:[e-mail]] 
> Sent: Friday, September 05, 2008 6:39 PM
> To: David Borman
> Cc: [e-mail]; [e-mail]; Mitesh Dalal (mdalal); 
> Anantha Ramaiah (ananth)
> Subject: Re: [tcpm] WGLC: draft-ietf-tcpm-tcpsecure-10.txt
> 
> On Thu, Aug 21, 2008 at 12:40:43PM -0500, David Borman wrote:
> > 
> > Wes and I would like to start the WG Last Call for:
> > 
> >  Title           : Improving TCP's Robustness to Blind In-Window  
> > Attacks'
> >  Author(s)       : A. Ramaiah, R. Stewart & M. Dalal
> >  Filename        : draft-ietf-tcpm-tcpsecure-10.txt
> >  Pages           : 27
> >  Date            : July 9, 2008
> >  Intended Status : Proposed Standard
> > 
> > http://www.ietf.org/internet-drafts/draft-ietf-tcpm-tcpsecure-10.txt
> > 
> > It is our understanding that all the feedback has been incorporated 
> > into this latest version and that there are no known outstanding 
> > issues with this document.
> > 
> > Please send feedback to the list, even if it is just a 
> "yes, go ahead 
> > and publish".
> > 
> > The WGLC will end on Friday, September 5, 2009.
> 
> Overall I think this is a good draft and should advance.  I 
> have comments, which I'll present in page order:
> 
> Probably the most technical comment I have is the objection 
> to the "There was no sound technical reasoning for choosing 
> the Data mitigation as MAY", which is on P15.
> 
> P2:
> The 4-tuple is called a socket pair in 793.  Even if you 
> don't want to adopt that terminology, it's probably worth 
> mentioning.  I also don't like "RST, SYN or DATA segment."  
> These are TCP segments with the RST or SYN flag set and I'd 
> describe them that way.  ONce you've had a chance to define 
> "RST segment" as a segment with the RST bit set, you can use 
> the short form.

Good point. Will do.

> 
> I'd delete "either" from "This can cause the connection to 
> either abort or possibly cause data corruption."  I'd 
> consider deleting "possibly" as well.

Makes sense. Will do.

> 
> P4:
> 
> "The TCP spoofing attacks, which are seen in the Internet 
> today," - I'd say "The off-path TCP spoofing attacks".  We're 
> still describing the problem, so let's be clear.  The next 
> paragraph starts with a description of the attacks this draft 
> cares about, but until then be precise.
> 
> "... making the odds of guessing correctly the 4-tuple..."  I 
> think "correctly guessing" is more common.  I'd actually 
> suggest "... making it much easier to guess the 4-tuple."

Ok. 

> 
> P5:
> 
> "One of the important things to note is that, for the attack 
> to succeed the RST..." -> delete the comma.
> 
> "A slight enhancement to the TCP's segment processing" -> 
> delete the "the".

Both points taken.

> 
> P6:
> 
> "Every application has control of a number of factors that 
> effect drastically the probability of a successful spoofing 
> attack."  You mean "affect" not "effect".  I'd also use 
> "drastically affect" rather than "affect drastically".  The 
> order of the adverb is a matter of style, using "affect" 
> changes the meaning of the sentence to the one you intend.

Cool. Will do.

> 
> On the figures in the "To successfully inject a spoofed packet..."
> paragraph:
> 
> 1. Use parens, not [].
> 2. Finish the math.  If you want to say 1/2 * 2^32 = 2^31 
> rather than just dropping 2^31 that's fine, but we're going 
> to be comparing these requirements, so don't hide the 
> results.  "[SITW] shows that the mean number of tries needed 
> to inject a RST command is (2^31/window) rather than the 2^31 
> assumed before."

Ok.

> 
> 
> P7:
> 
> "...please refer to draft [RFC4953]" -> "please refer to [RFC4953]"
> It's not a draft any more.

Ok.

> 
> P8:
> 
> Why use 1) and 2) and A), B), C)?  Pick one.
> 
> "The previous text, quoted from [RFC0793], woould thus 
> become:" -> That sounds a lot like we're mandating the change 
> rather than it being a SHOULD.  Do we really need this 
> restatement at all?

Hmm.. I thought it was more clearer that way. 

> 
> P13:
> 
> "...so the chances of successfully injecting data into a 
> connection are
> 1 in (2^32/RCV.WND *2)."  In describing the RST attacks, we 
> spoke in terms of mean number of tries, and I'd be consistent 
> here.  Similarly I'd do the math all the way: " ... so the 
> mean number of tries needed to inject data successfully is  
> 2*2^32/RWND = 2^33/RCV.WND."

Sure, it is good to be consistent. But the math which you depict above
seems incorrect to me. What was said earlier in the document is
(referenced from Watson's paper ) is :

============
   "[SITW] demonstrated that
   this assumption was incorrect and that instead of [1/2 * 2^32]
   packets (assuming a random distribution) [1/2 * (2^32/window)]
   packets is required.

   Substituting numbers into this formula we see that for a window size
   of 32,768, an average of 65,536 packets would need to be transmitted
   in order to "spoof" a TCP segment that would be acceptable to a TCP
   receiver.  A window size of 65,535 reduces this even further to
   32,768 packets.  At today's access bandwidths an attack of that size
   is feasible.

===============

So for data injection, this should become 2^32/RCV.WND. (i.e, [1/2 *
2^32 *2]/window )


> 
> This section used a) and b) instead of either A) and B) or 1) 
> and 2) (used earlier for the same purpose). Again, pick one.

Ok.

> 
> P15:
> 
> "There was no strong technical reasoning for choosing the 
> Data mitigation as MAY."
> First, everywhere else the you use "DATA", I'd use it here.

Sure.

> 
> Second, if I failed to make the case before, let me make it 
> now.  The DATA injection is 4 times less likely than either 
> the RST or the SYN attacks (2^33 vs 2^31) and may be visible 

Hmm.. It should be actually twice less likely 2^32 versus 2^31, based on
above.


> to the application as garbled data.  If the attack is visible 
> an application can gracefully terminate the connection and 
> re-establish the conversation - unlike the RST/SYN cases.  As 
> the attack is harder to carry out, and a successful attack 
> may be easier to recover from, I recommend a MAY.  I don't 
> care if you repeat that argument, but I do believe that there 
> are sound technical reasons for the MAY, and I'd like to see 
> the sentence that started this comment elided.

I think the general understanding was that any malicious event
responsible for the connection to go down is a problem. So I am not sure
whether I can buy this argument. OTOH, if you are classifying the
strength based on how easy/difficuly it is to contruct the attack vector
and cause harm to the connection, and if that is what is the consensus,
then so be it. I can mention this in the document.

> 
> P16:
> 
> "Currently there is no known bad behavior that can be 
> attributed to the lack of ACK throttling, but as a general 
> principle, if ever invoked, something incorrect is occurring 
> and such a mechanism will act as a failsafe that protects 
> both the sender and the network." -> "While we have not 
> encountered a case where the lack of ACK throttling can be 
> exploited, as a fail-safe mechanism we recommend its use.  An 
> implementation may take an excessive number of invocations of 
> the throttling mechanism as an indication that network 
> conditions are unusual or hostile."

Your change looks elegant. Will do.

> 
> P18:
> 
> "...the middle box design does not comply to [RFC0793]."   No 
> middlebox
> complies with RFC793; I suggest "...the middle box is 
> generating packets a conformant TCP endpoint would not generate."

Ok.

> 
> P22:
> 
> "ACK throttling was introduced to this document bt combining 
> the suggestions from the tcpm working group."  That seems out 
> of place in "Contributors".  That section is to acknowledge 
> the efforts of individuals.

Makes sense.. Good eye!

> 
> P24:
> 
> Why are RFC's 4302 and 4303 normative?  And if they are why 
> isn't RFC2385?  They're all referred to as possible 
> mitigations.  My preference is making 4302 and 4303 
> non-normative, but it's very likely that I'm missing a rule here.

I am not sure about the rules here?  David/Wes what needs to be done
here ?

> 
> Quotation marks are misplaced on the Medina05 reference.

Ok.

> 
> The NISCC reference lacks a date.
> 
> P25:
> 
> Quotation marks are misplaced on the SITW reference.

This was put in long back.... I need to check.

I'll incorporate these comments along with others received in the next
(final, hopefully ?) version.

Once again, thanks for the detailed review.

-Anantha
> 
> --
> Ted Faber
> http://www.isi.edu/~faber           PGP: 
> http://www.isi.edu/~faber/pubkeys.asc
> Unexpected attachment on this mail? See 
> http://www.isi.edu/~faber/FAQ.html#SIG
> 
_______________________________________________
tcpm mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/tcpm