Search RPD Archives
[rpd] REPORT ON Appeal against the non-consensus determination on proposal AFPUB-2019-GEN-006-DRAFT02 (RPKI ROAs for Unallocated and Unassigned AFRINIC Address Space – Draft 2).
fhfrediani at gmail.com
Tue Jan 26 13:51:06 UTC 2021
I think that was a very good and detailed email.
There is however one point I differ on you: the decision taken by the AC
should be based on majority of votes of the members. Obviously they must
not take decisions based on personal opinions of the merit of the
proposal but on the facts and if all issues have been resolved.
Regarding the decisions between the 2 Co-Chairs I don't think the
consensus model applies there simply because there is not how to have a
consensus model applied between 2 persons. Either they both agree on
something or that cannot advance.
In resume: The consensus evaluation between 2 Co-Chair require both to
agree there was consensus. The consensus evaluation or analysis if the
Co-Chairs committed a mistake between all members of the AC is done via
majority and in this last case that doesn't exclude the eventual
necessity of explaining in details how every evaluation was taken for
transparency proposes. The consensus model applies only the community
while the proposals are being discussed and issues are being addressed.
On 26/01/2021 07:50, JORDI PALET MARTINEZ via RPD wrote:
> Hi Wafa, all,
> First of all, don’t take anything that I say personally, but in
> general I see a total failure of the Appeal Committee and lack of
> compliance with the PDP.
> Your judgment must be on the grounds of a correct decision of the chairs.
> In taking such decision the Appeal Committee must be based on facts,
> never on personal opinions (from the community or the chairs or the
> Appeal Committee itself). Being based on objective facts means
> checking if what the policy proposal said, what were the objections,
> and if those objections **are real**, not just “illusions” or “lack of
> knowledge” or “untrue” or “personal preferences”.
> If the Appeal Committee doesn’t have the right knowledge, as I already
> said I believe was the reason the chairs took the wrong decision, then
> they should ask for help to the staff or third parties.
> Any objection to a policy proposal must be duly justified and that
> justification not addressed by the authors or other community members.
> Any policy proposal that has objections, the objections MUST BE VALID,
> even if the objections come from 99% of the community. This is not
> democracy, is not number of votes or voices, is based on non-addressed
> objections. It is not based on untrue objections. None of the
> objections to this policy proposal were valid. They are mostly based
> on lack of sufficient knowledge, and never lack of knowledge can be a
> VALID reason. Again, not only the authors, but many other expert
> community members have confirmed that those objections are invalid.
> A policy proposal never can be based in “I don’t like it”. You need to
> state “I don’t like it because it breaks this RFC” (for example). And
> even in that cases authors can respond showing why the perception of
> “breaking this RFC” is wrong (so addressing the objection will nullify
> it). Policies are not based on personal preferences, but in what is
> the best **technically correct choice** for the community.
> Last but not least, the Appeal Committee seems to be working as a
> democratic body, which is wrong. ALL THE PDP is based in consensus
> approach. The Appeal Committee must also follow that approach,
> otherwise, it is breaking the ICP-2, which is the higher mandate of
> how the policy making process works. If 3 members of the Appeal
> Committee believe that the opposition was correct, they should
> **demonstrate with facts why** and this must be done using the
> responses provided by the authors and community to those objections.
> If 3 members of the Appeal Committee believe that any of the
> objections has not been addressed, they need to **demonstrate why**,
> taking in consideration the community and author responses, and those
> must be crystal clear in the report, which is not the case.
> The Appeal Committee must respond to the authors, in a consensus based
> approach, not a democratic one to all what the authors confirmed in
> the Appeal Document.
> Note also that there is a paragraph in the Appeal Report that
> completely kills the PDP and demonstrates that the Appeal Committee
> HAS NOT UNDERSTOOD THEIR JOB AT ALL:
> “The 3 members who observed significant opposition to the policy,
> however, also observed that it is the PDWG that builds consensus and
> decides whether issues of opposition are addressed to the satisfaction
> of the PDWG which is where the PDP requires that consensus is assessed
> by the Co-Chairs.”
> The PDP states clearly that the Appeal Committee need to review the
> chairs decision. If the chairs have considered as VALID objections
> that OBJECTIVELY ARE INVALID, it is clear that the Appeal Committee
> must declare the lack of consensus declaration is invalid, and
> consequently, the proposal reached consensus.
> Let’s go the details and I ask the Appeal Committee to respond to each
> of the objections included and refuted in the Appeal Document:
> 2.1.“a. Allowing resource holders to create AS0/ ROA will lead to an
> increase of even more invalid prefixes in the routing table”
> Following RFC6483, section 4 (“A ROA with a subject of AS 0 (AS 0 ROA)
> is an attestation by the holder of a prefix that the prefix described
> in the ROA, and any more specific prefix, should not be used in a
> routing context”) resource holders, as part of the RPKI system already
> can and actually do this (example IXPs), in fact they do. This has
> been explained several times, including my presentation at the PPM.
> The proposal is just adding light about actual facts regarding this
> aspect, not changing anything, so it *can’t be a valid objection for
> the policy proposal*.
> 2.2.“b. Revocation time of AS0 state, and the time for new allocation
> doesn’t match”
> This is not true, again a misunderstanding about how RPKI works. The
> authors and other several community experts have discussed this in the
> list. If you get number resources from AFRINIC, normally you don’t
> announce them in minutes, or hours, or even days. There is some work
> to do in your network, you need to do changes in systems and routers,
> and this takes hours, and normally you can’t “touch” systems during
> the day, but only in “maintenance windows” in the night. That means
> that if AFRINIC revokes an AS0 certificate, it will be done in a few
> minutes during the day. So even if the worldwide caches take hours to
> see that, there is no negative impact.
> In addition to that, this it can be improved thru implementation, as I
> already explained also in the list. The staff could tentatively
> release from the AS0 the resources that they plan to allocate once a
> week or every couple of days, etc. For example, when they are
> processing a request, and they are pending on final documentation, the
> RSA signature for new members, or the review with the member of the
> justified need. Many other examples can be provided about how to do
> it. The proposal doesn’t go into any of those details, because the
> understanding is that those are too depth operational, and in fact the
> staff could decide an approach during the implementation, and based on
> experience improve it afterwards.
> The conclusion is that there is no such “matching”, neither
> “unmatching”, so this *can’t be taken as a valid objection for the
> 2.3.“c. Other RIRs don’t have a similar the policy therefore, it can
> not be effective”
> All the policies have different discussions in different RIRs at
> different times. This policy is already available (reached consensus
> and implemented) in APNIC and LACNIC (reached consensus, being
> implemented). There is work being done in ARIN and RIPE (the first
> proposal was not accepted), not yet public. So, this is untrue if you
> look at all the RIRs.
> The effectivity of a policy is not only related to how many RIRs
> implement it. In this case, any RIR having this policy is actually
> stronger than the other RIRs not having it, in terms of routing
> security. It shows the commitment of that RIR about the RPKI usage
> with all its possibilities. It facilitates the operators in the region
> and outside the region to identify in a simpler and automated way,
> what prefixes should not be in the routing tables and consequently
> allow them in an opt-in basis, to discard them. So, it is in the other
> way around, any RIR with this policy could be said that it is more
> “effective” (even if it is not probably the right wording for this
> topic) that the others. We should rather say that “a RIR with this
> policy is offering a more secure view of their routing information”.
> In addition to that, there are policies in AFRINIC which aren’t
> available in other RIRs. That, clearly, doesn’t make them invalid (or
> in other words, this is an invalid objection – is good that all RIRs
> do the same, but is not always the case, or not at the same time),
> clearly this shows that this *can’t be taken as a valid objection
> against this policy proposal*.
> 2.4.“d. This will become a uniform policy if it is not globally
> implemented, which causes additional stress“
> This is almost a duplicate of the previous one. Absolutely not. We can
> add that the way we suggest the staff, and they confirmed, with an
> independent TAL protects, as intended by the proposal, the resources
> of the RIR implementing it, not creating any issues in what is done in
> other RIRs to any operator, so it *can’t be taken as a valid objection
> against this policy proposal*.
> It is difficult to understand what it means “additional stress” in
> this context, but clearly, it will be in the other way around. As more
> RIRs implement it, less manual work in terms of filtering, is needed
> to be done by operators, if they opt to use the AS0 ROA service from
> the RIRs that have implemented it. So, it is not correct and thus,
> *not a valid objection*.
> If the question is about if this policy should be a Global Policy, the
> response has also been provided in the discussion. Ideally, a Global
> Policy will be only able to protect the IANA unallocated resources,
> but not the resources that IANA already allocated to each RIR. In
> fact, I’m already working (when time permits it will be made public)
> in a Global Policy for that, but this is irrelevant versus having a
> policy at every RIR (or a few of them), so again, objectively *not a
> valid objection*.
> 2.5.“e. Validity period: if members decide to implement it, is it
> not better to recover the space if it is kept unused for too long?”
> This doesn’t make sense, at least not as worded. This is not about
> recovering space, no relation. It is the unused space hold by AFRINIC,
> regardless of if it was never allocated/assigned, returned by members,
> or recovered by AFRINIC. Once more, *not a valid objection*.
> 2.6.“f. How do we revoke the ROA? How long does it take to revoke it
> (chain/ refreshing )?”
> This looks the same as 2.2 above. It doesn’t matter in practice, if it
> takes minutes or hours or even days. Community and staff provided some
> facts about this, just to cite a couple of them:
> 2.7.“g. What happens if AFRINIC accidentally issues a ROA for an
> address in error?”
> What happens if AFRINIC accidentally issues a ROA without an address
> already allocated to members?
> Exactly the same if the existing RPKI fails, and that’s why there are
> monitoring systems in place and as reported by the staff impact
> analysis, this proposal will ensure that the monitoring is improved,
> so it is actually helping on the right direction, not in the other way
> Further to that, because the systems of operators have caches, it is
> all depending (for the existing TAL and for the new one implemented
> with this proposal) on how much time it takes to AFRINIC to resolve
> the error and the specific configuration of the operators and if they
> actually drop invalid prefixes or they only create alerts, trigger
> some processes, etc. Note that RPKI doesn’t force the operators to
> drop the prefixes even if they are using RPKI, there are different
> ways to take advantage of this.
> This proposal doesn't change that, it is provided as an opt-in service
> and consequently it is *not a valid objection*.
> 2.8.“h. It also might affect the neighbours and involves monitoring of
> unallocated spaces”
> It is not clear if neighbours here is referring to BGP peering ones.
> The same monitoring that right now is done AFRINIC for
> unallocated/unassigned spaces could be improved with this proposal.
> AFRINIC already, today, needs to make sure that they get alerts if
> unallocated/unassigned space appears in the routing tables, because
> that may imply that a member may be violating the RSA, bylaws,
> policies, etc.
> Exactly the same as for operators connected to Internet with BGP. The
> proposal allows them, as an opt-in service, to improve if they wish,
> the automation of all that, and to use the service in the way they
> decide. The proposal is not forcing operators any specific usage for
> the service, it is entirely at their own decision/discretion.
> This proposal ensures that the service is improved so, hijacking of
> unused space is less prone to occur, that’s the purpose of the
> proposal and RPKI, increase the routing security, without making it
> mandatory for any operator. Consequently, once more, this *can’t be
> considered a valid objection*.
> 2.9.“i. Possibility of it being used against a member who is yet to
> pay dues”
> According to AFRINIC bylaws and RSA, AFRINIC has the *obligation* to
> avoid members not paying to stop using the resources, so they are
> available to other members.
> It will be unfair and discriminatory to other members not doing so,
> and that’s the reason, even if we don’t have this proposal, AFRINIC
> could at any time, following the bylaws and RSA, do whatever actions,
> including legal and technical ones, to make sure that unallocated, or
> unassigned, or returned, or recovered resources are not used. As part
> of those actions, AFRINIC could even ask in courts to stop routing
> those resources, even to other operators. It is AFRINIC duty,
> practically will probably not make sense in terms of the cost (unless
> a major hijacking of AFRINIC resources occurs). Most probably just the
> cooperation among operators, without any legal requirement, will make
> that happen. So, this proposal doesn’t change that in the sense of
> adding to AFRINIC any new prerogative because already have that right
> and duty regarding the responsible use of the resources only to the
> allocated/assigned parties and in compliance with the legal bindings.
> To further explain this, if a member decides to stop paying, AFRINIC,
> following legal bindings, will follow a procedure to try to fix it,
> and if it doesn’t succeed, will remove whois data (which in turn will
> cause the removal of route objects that depend on them), RDNS (which
> means the address space becomes in general unusable), etc.
> Clearly, once more, this *can’t be considered a valid objection, on
> the other way around is a fundamental AFRINIC’s right and duty*.
> I urge you to respond to each of those objections, accepted by the
> chairs to declare the lack of consensus, that the authors and other
> community members DEMONSTRATED with OBJECTIVE information, are invalid.
> Again, please, Appeal Committee members, respond OBJECTIVELY AND BASED
> ON FACTS, NOT PERSONAL PREFERENCES. The report MUST contain detailed
> demonstration of why the Appeal Committee (not individual members) say
> each of those objections has not been addressed, while the authors and
> community believe otherwise.
> This is what we expect from an Appeal Committee, to OBJECTIVELY review
> what the chairs objserved, when the Appeal Document clearly
> demonstrated that it is invalid and consequently the chairs took a
> wrong decision, based on personal preferences of community members or
> lack of knowledge, or other not objective or untrue facts.
> El 22/1/21 13:59, "wafa Dahmani" <wafatn7604 at gmail.com
> <mailto:wafatn7604 at gmail.com>> escribió:
> Dear Community,
> This is to inform you that the Report on Appeal against the
> non-consensus determination on proposal AFPUB-2019-GEN-006-DRAFT02
> (RPKI ROAs for Unallocated and Unassigned AFRINIC Address Space –
> Draft 2) and the minutes have been published following the links below:
> Best Regards
> Wafa Dahmani
> Chair of the Appeal Committee
> _______________________________________________ RPD mailing list
> RPD at afrinic.net https://lists.afrinic.net/mailman/listinfo/rpd
> IPv4 is over
> Are you ready for the new Internet ?
> The IPv6 Company
> This electronic message contains information which may be privileged
> or confidential. The information is intended to be for the exclusive
> use of the individual(s) named above and further non-explicilty
> authorized disclosure, copying, distribution or use of the contents of
> this information, even if partially, including attached files, is
> strictly prohibited and will be considered a criminal offense. If you
> are not the intended recipient be aware that any disclosure, copying,
> distribution or use of the contents of this information, even if
> partially, including attached files, is strictly prohibited, will be
> considered a criminal offense, so you must reply to the original
> sender to inform about this communication and delete it.
> RPD mailing list
> RPD at afrinic.net
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the RPD