Re: [hiprg] HIP experiment report comment on opportunistic mode

["Henderson, Thomas R" <[e-mail]>]

> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]] On Behalf
> Of Miika Komu
> Sent: Tuesday, December 06, 2011 7:09 AM
> To: [e-mail]
> Subject: Re: [hiprg] HIP experiment report comment on opportunistic
> mode
> 
> Hi,
> 
> On 12/06/2011 10:21 AM, Tobias Heer wrote:
> > Hi,
> >
> > Am 06.12.2011 um 09:01 schrieb Henderson, Thomas R:
> >
> >> I would like to respond to one of Stephen Farrell's comment on the
> >> HIP experiment report:
> >>
> >> http://datatracker.ietf.org/doc/draft-irtf-hip-experiment/ballot/
> >>
> >> The comment is:
> >>
> >> " - I wondered what were the "controversial experiences" on p10.
> >> Seems a shame to tease the reader like that."
> >>
> >> The relevant section of text is:
> >>
> >> In the context of the HIPL project, the opportunistic mode has
> >> been successfully applied at the HIP layer for service
> >> registration [RFC5203].  However, there are controversial
> >> experiences on applying opportunistic mode at the application layer
> >> for legacy software. HIP4BSD implemented opportunistic mode
> >> successfully with small modifications to the FreeBSD socket layer
> >> to support opportunistic mode.
> >>
> >> Could someone elaborate on the controversial experience (and
> >> perhaps provide a reference)?  Note that elsewhere in the report
> >> (section 2.3.2), the disadvantages and "leap of faith" aspects of
> >> opportunistic mode are elaborated on, so I'm wondering whether the
> >> reference to controversial experiences goes beyond the
> >> disadvantages already listed in section 2.3.2 (or whether we could
> >> instead strike those words from the draft and refer back to that
> >> section).
> >>
> > As far as I can tell, we in Aachen have not used the opportunistic
> > mode extensively. Miika seemed to have used it/struggled with it
> > quite a bit. Maybe he has some comments on this.
> 
> we implemented opportunistic HIP mode at two different levels in Linux:
> 
> i. As an intercepting SHIM library between the application and the libc
> socket calls implementations using LD_PRELOAD
> ii. As an intercepting SHIM layer between transport and network layers
> using iptables
> 
> While the implementation was a success, it was not far from ready for
> production use and we decided to remove the data plane part from the
> implementation. However, the control planet part (opportunistic base
> exchange) is still left and can be used e.g. for registering for
> rendezvous.
> 
> So I guess the "controversial" means perhaps that the data plane
> processing was more difficult to implement on Linux than on BSD, I
> suggest rewriting something along these lines or just referencing the
> paper for details (you have the reference already in the draft).

OK, I will delete the sentence mentioning controversy, and add a sentence at the end of the paragraph that states:  "However, the Linux implementation was more challenging as described below."  The following two paragraphs already go into the details.

> 
> Btw, I would appreciate if you could reference also the following paper
> in the report:
> 
> Kristiina Karvonen, Miika Komu and Andrei Gurtov, Usable Security
> Management with Host Identity Protocol, published in The 7th ACS/IEEE
> International Conference on Computer Systems and Applications (AICCSA-
> 2009)
> 
> It supplements opportunistic and normal HIP experiments with usability
> test results. We also report experiences from using a graphical end-
> host
> firewall. I would kindly ask a reference, e.g., to section 2.3.2:
> 
> "...or by prompting the user using a graphical interface to explicitly
> accept the connection [REF]."

Thanks, I was not aware of this interesting paper.  Will be added to the -15 draft.

- Tom