[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: PKI Resource Discovery - Proposal for a new Working Item



Anders Rundgren wrote:
Hi Max,

Hello Anders,

In case you find that there is limited interest in PRQP, I encourage
you to explore other avenues in this space.

Well, I am always open to investigate other possibilities. One thing about
PRQP is that some discovery system like this would enable so many new
possibilities in organizing PKIs (and managing them), that I think it
would really ease management of PKIs and adoption of certificates.

As the OpenCA Program Manager, I guess you are aware of the fact
that on-line provisioning of certificates is not fully standardized?

Right, well.. it should be, but it is not yet. Practice is quite far
from being standardized. One thing that really strikes me is that
still today, you mostly need a browser to interact with CAs.

I am also working at an open source PKI-enabling library - which will
be the core of the OpenCA-Ng (Next Generation - where we have to include
support for on-line provisioning of certificates in such a way that
it is easy for the developer to support interactions with the CA
(my assumption here is that, if it is easy for the developer, the
interface will be also easier for the user as the developer will not
simply transfer all the options onto the user because of lack of
knowledge about PKIs).

One could consider Xenroll a standard since it is supported by 80%
of  the browsers used in PCs.  However,  Xenroll is not supported by
more than a tiny faction of mobile browsers.  The latter is an
interesting target given the 3Bn+ users that will most likely use mobile
phones as their primary, always connected Internet channel.

Right...

Theoretically one could distribute keys in SIM cards, but for
practical reasons like operator lock, limited storage, and poor
processing capability, TPMs as defined by TrustedComputingGroup
looks like a better candidate for the universal mobile "key-ring".

Well.. if we do not solve the resource discovery and interoperability
between PKIs, then TPMs will always be a nice but unused piece of
HW. I have looked at the TCG work, it is a nice effort, but it is
far from being even usable in closed and controlled environments
(at least for its initial "purpose", i.e., remote attestation).

The only usage of TPM, today, is to provide runtime memory protection
or key storage/usage. And I think it is an easier way to provide some
kind of HW protection for keys.. although... new keys are really stored
in the FS of the hosting machine(..), it is a start...

Various radio-technologies potentially also open these keys for
desktop usage where the phone becomes a "security device" including
[..]
XML protocol giving a uniform user experience and an easier-to-secure
implementation (APIs can be used in many ways, while strictly defined
XML schema-based protocols give little room for misusage).

Well, in general I am not really a fan of XML + Schema usage for
certificates. Besides the fact that I love XML.. it is easy for the
user... but when it comes to certificates (especially if you take in
consideration small devices and the possibility to have pkis integrated
into them -- e.g., sensors, mobile phones, etc.. ) I would stick with
a more compact message format (DER). This, mainly, because one of the
nice features of XML is the possibility to validate the messages by
using schemas, i.e., the application is freed from the need to check
the message syntax. Anyhow, to do so, you have to provide schemas and
the device should also be capable of verifying the message against the
schema - requiring quite a lot of computational power.

That is why, now, I think XML is not the best choice for PKI operational
protocols. It can be a choice when considering more high-level applications,
but this is just my opinion.

===============================================
Anyway, I am currently in a _v_e_r_y_ early stage of addressing this
topic and would not mind cooperation with other knowledgeable people.
===============================================

I guess this is the right place to ask for collaboration. We really need
to discuss some of the features that would help PKIs to provide more inter
operable services :) It would be interesting to discuss the topic in detail
and if you can come up with a proposal... one thing we need is to have a
description of the current practices and various options (standardized and
non standardized) we have - so we do not duplicate existing work. I guess
this could be a good starting point.

Regarding PRQP, I still feel a little bit puzzled regarding the
resources it is supposed to discover.  A few examples would not hurt.

An example tied to your idea would be the following.

As you said different CAs support different protocols/procedures to provide
certificates to its users. If PRQP is used by a CA, a client could ask which
services are provided -- and the CA could reply with a list of services
supported, e.g., if CMS is supported, an URL could be provided for that.
If web (e.g., Xenroll) is supported, another URL could be provided for
that as well. The client will then contact the URL that is supported.

Another scenario where PRQP could be very useful is for service rollover.
For example if a CA starts with providing CRLs and then it wants to provide
OCSP only because CRLs are too big (e.g., DoD problems with CRLs), the
PRPQ responder can dynamically redirect clients from one service to another.

Another scenario could be adding new servers to existing ones to provide
fall back servers to clients without requiring configuring round-robin or
other more complicated load-balancing DNS-based service.

Well, as I said before, PRQP could really open up new possibilities, that
is the most interesting thing about it, I guess.

Later,
Max

--

Best Regards,

	Massimiliano Pala

--o------------------------------------------------------------------------
Massimiliano Pala [OpenCA Project Manager]            pala@xxxxxxxxxxxxxxxx
                                                 project.manager@xxxxxxxxxx

Dartmouth Computer Science Dept               Home Phone: +1 (603) 397-3883
PKI/Trust - Office 063                        Work Phone: +1 (603) 646-9179
--o------------------------------------------------------------------------

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature