Re: [dhcwg] [Last-Call] Secdir last call review of draft-ietf-dhc-dhcpv6-pd-relay-requirements-04

[Christian Huitema <[e-mail]>]

I do not think this is a new issue that should be fixed in this document. I have written that a couple times already. I do think that it would be nice to work on another document describing best practices to secure DHCP services.

-- Christian Huitema 

> On Dec 7, 2020, at 6:01 PM, Bernie Volz (volz) <[e-mail]> wrote:
> 
>  There really aren’t any. Clients can generate different client-id and request prefix after prefix. 
> 
> However, as this document points out these relays usually have limits to what it may track (and likely it would drop those packets from reaching client).
> 
> In some deployments, the number of prefixes allowed behind a particular relay is limited (at server or relay). Other mitigations may be shorter leases as then it comes down to how many can be requested during that time.
> 
> The main question is what does this benefit anyone? It is a DoS but in general it has limited impact as prefixes tend to be topological, so it isn’t like you could assign all of the prefixes an ISP has — just what is allowed on that link.
> 
> Why do you think this is a new issue that needs to be fixed in this document?
> 
> In my experience we have not seen these kinds of attacks as they aren’t very useful. And it has been a dhcp issue since dhcpv4 (addresses were much more scarce).
> 
> We tried securing dhcpv6...but it isn’t an easy problem to solve.
> 
> - Bernie
> 
>>> On Dec 7, 2020, at 8:06 PM, Christian Huitema <[e-mail]> wrote:
>>> 
>> 
>> 
>> 
>>> On 12/7/2020 4:31 AM, Bernie Volz (volz) wrote:
>>> FYI:
>>> 
>>>>> I understand that solutions like RA
>>>>> Guard will in practice provide some protection, but the use of these solutions are
>>>>> not discussed in RFC 8213. The DHCP WG might want to address that.
>>> 
>>> RFC8415’s security considerations is rather extensive and includes reference to many techniques to reduce the issues. 8213 was written while 8415 was under development.
>> In the context of the draft, I am concerned in particular with the "resource-exhaustion" DoS attack, through exhaustion of delegatable prefixes. The attack is mentioned in the security section of 8415, but I have not seen the proposed mitigation.
>> 
>> -- Christian Huitema
>> 
>>