Re: [Lake] proposed scoping text

[Christopher Wood <[e-mail]>]

On Mon, Apr 27, 2020, at 2:23 AM, Göran Selander wrote:
> Hi Christopher,
> 
> As you remember, the scope of this work have been discussed (outside 
> ACE) since Jan 2019. Support for PKI and certificates has been included 
> throughout the different steps, the virtual Secdispatch interim March 
> 2019, the LAKE BoF July 2019, etc., each occasion followed by a 
> consensus call with large support. Removing all kind of PKI support 
> from the scope would be a drastic last minute change.

I'm aware, but I think it's the right thing to do in the near term. 

> The use of certificates is important for efficient trust management in 
> deployments with large numbers devices, and is one of the driving 
> forces for moving away from PSK-based IoT deployments. The number of 
> IoT devices is projected to surpass the number of web servers world 
> wide by orders of magnitude. A common building automation deployment 
> can have thousands of devices from several manufactures. If this is not 
> the setting where we want to use a PKI, then what is? 

I can't comment on whether or not this makes sense. I don't work in the space. All I'm suggesting is to focus on small cases first. 

I'll note also that it's *possible* to have your cake and eat it too, as demonstrated by Tailscale [https://tailscale.com/blog/how-tailscale-works/]. We could design an AKE that works with only public keys, or even PSKs for that matter, and move everything else to the application layer. To me, this demonstrates the certificates need not be baked into the protocol to solve the use case you (may?) have in mind.

> If we want the IETF to provide a relevant contribution to securing 
> constrained IoT we cannot exclude device certificates. Instead we 
> should design the AKE so that these deployments can use certificates in 
> a simple and efficient way. Referencing certificates instead of 
> transporting them over constrained links is a key component here.
> > I think including certificates complicates the story and will make 
> > progress on an efficient AKE
> 
> I am not convinced by the "complication" argument because it is hard to 
> measure at this stage. I think the scope should to be defined by what 
> we need to solve and what we believe we can solve efficiently, as was 
> motivated in the proposal.
> 
> > I thought Ben's suggestion to focus only on PSK and RPK modes to start 
> > was fine, though this recent proposal seems to also support certificates, '
> > albeit only by "reference." 
> 
> Please read the first line in the proposal again. The proposed scope is 
> RPK + certificate by reference, not PSK + RPK + certificate by 
> reference. 

This is *your* proposal, not Ben's. I'm referring to (and prefer) Ben's suggestion, for the reasons stated in my email. 

> So, this is in fact good news for those who are concerned 
> about complications, since removing the PSK based authentication 
> protocol is a significant simplification of design, specification, 
> analysis, implementation and testing. 

Are you suggesting that the holistic design and implementation of an AKE that uses certificates is less complicated than use of PSKs?

> But, again, I think we should 
> refrain from the "complication" argument and instead focus on what we 
> need to solve (and to get started!). 
> 
> In contrast to removing certificates from the AKE, removing PSK seems 
> not to notably impact the desirable scenarios where this can be 
> applied, since RPK by reference can perform as good as PSK. 
> Furthermore, I interpreted Ben's mentioning of PSK and RPK as just an 
> example, based on the discussion we had at the latest LAKE virtual 
> interim about the cases where a candidate protocol can perform 
> optimally. Not that this subset necessarily was the subset we should 
> restrict to. 
> 
> > it assumes some type of PKI in which these certificates exist and through 
> > which endpoints handle tasks such as revocation and transparency. 
> > That may impose requirements on the AKE. TLS had to carve out 
> > extension support for CT and OCSP. Is that something we want to take 
> > on when starting down the path towards an efficient AKE?
> 
> The efficiency of the AKE need not be restricted by allowing validation 
> of certificates. This discussion really belongs to the design phase but 
> starting it anyway, there are two cases:

I think the important bit here is that the AKE (TLS) had to be extended to support these emerging technologies after-the-fact. This suggests the AKE should minimally be extensible if things such as revocation are not in scope for an initial certificate-focused design. It may have other implications not yet known, too. Have we carved out enough space for these sorts of things? I'm not sure.

Best,
Chris

> 1. The endpoint receiving the certificate reference can retrieve 
> information about the validity of the certificate over a 
> non-constrained link, see e.g. draft-selander-ace-ake-authz. In this 
> case no validation information needs to be carried in the AKE. The 
> validation is performed by the endpoint according to policy and is 
> essentially decoupled from the AKE.
> 
> 2. The endpoint has access to the certificate but not timely validation 
> information. For example if the device is built-to-order and 
> pre-provisioned with the target network certificate(s). In this case 
> some validation information may need to be carried in the AKE. 
> 
> I think we should keep also case 2 in the initial scope, at least to 
> allow us to determine during the design phase if there is a simple 
> lightweight solution that matches existing standards and tools, or else 
> leave it for later. I committed one change to the PR to this point:
> 
> https://github.com/lake-wg/reqs/commit/91d32c2
> 
> 
> Göran
> 
> 
> 
> On 2020-04-25, 02:17, "Lake on behalf of Christopher Wood" 
> <[e-mail] on behalf of [e-mail]> wrote:
> 
>     Apologies for the delay, all! I thought Ben's suggestion to focus 
> only on PSK and RPK modes to start was fine, though this recent 
> proposal seems to also support certificates, albeit only by 
> "reference." While this seems like a small addition at face value -- 
> what's different from a certificate reference with size equal to an 
> RPK? -- I think it changes the shape of the desired AKE in ways that 
> are perhaps unintentional. For example, it assumes some type of PKI in 
> which these certificates exist and through which endpoints handle tasks 
> such as revocation and transparency. That may impose requirements on 
> the AKE. TLS had to carve out extension support for CT and OCSP. Is 
> that something we want to take on when starting down the path towards 
> an efficient AKE? I'm not sure. Moreover, it does seem like the 
> environment in which this type of AKE exists and would be used is 
> different from one that only supports PSKs and RPKs. (The latter being 
> an environment with less dependencies.)
> 
>     Overall, while I appreciate the attempt to clarify scope, I think 
> including certificates complicates the story and will make progress on 
> an efficient AKE -- the goal of this group -- more difficult. I 
> strongly suggest we limit this to PSKs and RPKs only.
> 
>     Best,
>     Chris
> 
>     On Mon, Apr 20, 2020, at 5:25 AM, Mališa Vučinić wrote:
>     > Hi all,
>     > 
>     > This is a kind reminder to take a look at the proposed PR and 
> provide 
>     > feedback, especially if you disagree with the proposed text. With 
> the 
>     > remaining issues on the requirements documents resolved, we would 
> like 
>     > to converge on the best way forward for the working group, so 
> please 
>     > provide your comments by the end of this week, Friday the 24th.
>     > 
>     > Mališa
>     > 
>     > > On 15 Apr 2020, at 14:19, Göran Selander 
> <goran.selander=[e-mail]> wrote:
>     > > 
>     > > Hi,
>     > > 
>     > > Here is a proposal how to formulate the scoping in the 
> requirements draft:
>     > > https://github.com/lake-wg/reqs/pull/26
>     > > 
>     > > Göran
>     > > 
>     > > 
>     > > On 2020-04-09, 12:09, "Lake on behalf of Göran Selander" 
> <[e-mail] on behalf of 
> goran.selander=[e-mail]> wrote:
>     > > 
>     > >    All,
>     > > 
>     > >    TL;DR: I propose LAKE to initially focus on RPK, and 
> certificate by reference.
>     > > 
>     > >    =========
>     > > 
>     > >    Apologies for a long mail. Comments on Ben's proposal for a 
> way forward at the end, but first we need to take a step back and try 
> to understand what we are discussing.
>     > > 
>     > >    We are working on requirements for an AKE for OSCORE. We 
> require a lightweight AKE, since OSCORE may be applied in constrained 
> IoT applications. Now there is a discussion about restricting the 
> functionality of LAKE to address only the most lightweight scenarios, 
> so that another AKE should be used in non-lightweight settings. While 
> this may be an economy of standardization, this does not make sense for 
> many reasons.
>     > > 
>     > >    * Note that OSCORE is expected to be applied in both in 
> constrained and non-constrained environments. OSCORE works whereever 
> CoAP does, and CoAP is expected to be applied in PSK, RPK and 
> certificate settings.
>     > > 
>     > >    * Implementation-wise OSCORE is more or less integrated with 
> CoAP, because it reuses CoAP processing. A combined OSCORE/AKE 
> implementation will thus be more or less integrated in CoAP. So how 
> should a developer handle the situation that there is a specification 
> of AKE1 addressing part of the credential space and another completely 
> incompatible specification AKE2 addressing another part? 
>     > > 
>     > >    * OSCORE is designed for end-to-end security from 
> constrained device to cloud. To be sure to support OSCORE based 
> security you need to implement both AKE1 and AKE2.
>     > > 
>     > >    * How could it be better security-wise to have two different 
> AKEs for one security protocol? Twice the number of security analyses, 
> etc. are needed. 
>     > > 
>     > >    * Many companies want to move away from symmetric-key based 
> IoT deployments to public key based, ideally to device certificates. If 
> their current technology does not support transport of certificates, 
> but maybe will later,  should they implement AKE1 or AKE2?
>     > > 
>     > >    * etc.
>     > > 
>     > >    Summarizing: At some point in time there needs to be one 
> single AKE for OSCORE that is sufficiently lightweight to perform well 
> in the most constrained settings but also has the functionality to 
> protect CoAP for whatever settings it is expected to be deployed in, 
> and that involves PSK, RPK and certificates.
>     > > 
>     > >    =======================================
>     > > 
>     > >    Having said this, it may still be relevant to discuss which 
> cases are most important right now, bearing in mind that this scoping 
> should not prevent later adding the remaining components.  
>     > > 
>     > >    One way to prioritize the scope of the first work on the AKE 
> is to address a) the needs from current use cases, and b) the potential 
> to provide something useful that is not already available:
>     > > 
>     > >    a) There is a general request to deploy a lightweight AKE 
> with OSCORE targeting general CoAP use cases 
>     > >    i.e. based on PSK, RPK as well as certificates. More 
> specific work items I'm aware of include RPK-based solutions (e.g. LoRa 
> Alliance) and mixed RPK/certificate based solution (e.g. 
> draft-selander-ace-ake-authz). As mentioned in my previous email, a 
> PKI-based solution is necessary to handle IoT deployments with a large 
> number of devices and many manufacturers, but the certificate does not 
> necessarily have to be transported.
>     > > 
>     > >    b) It is possible to do PSK and RPK by reference for the 
> most constrained benchmark (1,1,1). RPK by value and certificate by 
> reference can also be very lightweight, but may require more than one 
> fragment per message. However, in case of transporting a chain of X.509 
> certificates, or even a single X.509 certificate, it is not clear that 
> a lightweight AKE performs significantly better. 
>     > > 
>     > >    The intersection of a) and b) correspond to where there is 
> an active interest in a solution and where the solution can be made 
> lightweight: RPK (by reference and value) and certificate by reference. 
>     > > 
>     > >    This restricted scope leaves out the most lightweight and 
> the least lightweight options. Removing symmetric key based 
> authentication from first scope is a significant simplification since 
> this is a different protocol in many respects compared to the 
> asymmetric version. Moreover, since PSK and RPK by reference can be 
> handled with the same transmission cost, there is less motivation for 
> using PSK ECDHE as an intermediate deployment step.
>     > > 
>     > >    Now for Ben's proposal:
>     > > 
>     > >    On 2020-04-08, 17:41, "Lake on behalf of Stephen Farrell" 
> <[e-mail] on behalf of [e-mail]> wrote:
>     > > 
>     > >        Ben's proposal is:
>     > > 
>     > >        "
>     > >        Strip down the requirements document a lot, to have a
>     > >        qualitative sense of "these are the combinations of 
> crypto
>     > >        primitives that we consider important *right now*" (e.g.,
>     > >        PSK and RPK). Have an acknowledgment that extensibility 
> is
>     > >        inevitable, but disclaim that we are focusing on getting 
> it
>     > >        right for these narrow cases right now, and if someone
>     > >        wants to do a broader case in the future, then more
>     > >        analysis will need to be done at that time.  
>     > > 
>     > >        But for now,
>     > >        we focus on getting RPK and PSK into:
>     > > 
>     > >         3 flights, 51-byte messages, and do that well.
>     > > 
>     > >    [GS] Focus on RPK (by value and by reference) and 
> certificate by reference, and do that well. For RPK by reference: 3 
> flights in (1,1,1) fragments. For RPK by value and certificate by 
> reference the number of fragments should be kept at a minimum.
>     > > 
>     > > 
>     > >        If we end up with protocol X that does RPK well and
>     > >        someone has it deployed but wants to expand their
>     > >        deployment to also use certificates, they can extend
>     > >        protocol X to do that and have a fairly homogeneous
>     > >        deployment, vs. having to have a mix of protocol X
>     > >        and TLS.  It may well be that TLS would do fine for
>     > >        their extended use case, but not the original use
>     > >        case, and having a homogeneous deployment is valuable."
>     > > 
>     > >    [GS] Agree.
>     > > 
>     > >    Göran 
>     > > 
>     > > 
>     > > 
>     > >    -- 
>     > >    Lake mailing list
>     > >    [e-mail]
>     > >    https://www.ietf.org/mailman/listinfo/lake
>     > > 
>     > > 
>     > > -- 
>     > > Lake mailing list
>     > > [e-mail]
>     > > https://www.ietf.org/mailman/listinfo/lake
>     > 
>     > -- 
>     > Lake mailing list
>     > [e-mail]
>     > https://www.ietf.org/mailman/listinfo/lake
>     >
> 
>     -- 
>     Lake mailing list
>     [e-mail]
>     https://www.ietf.org/mailman/listinfo/lake
> 
>