Status of This Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is athttps://datatracker.ietf.org/drafts/current/.¶

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶

This Internet-Draft will expire on 23 April 2022.¶

Copyright Notice

Copyright (c) 2021 IETF Trust and the persons identified as the document authors. All rights reserved.¶

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶

Table of Contents

1.Introduction

Performing DNSKEY[RFC4035] algorithm transitions with DNSSEC[RFC4033] signing is unfortunately challenging to get right inpractice without decent tooling support. This document weighs thecorrect, completely secure way of rolling keys against an alternate,significantly simplified, method that takes a zone through an insecurestate.¶

Section 4.1.4 of[RFC6781] describes the necessary steps requiredwhen a new signing key is published for a zone that uses a differentsigning algorithm than the currently published keys. These are thesteps that MUST be followed when zone owners wish to haveuninterrupted DNSSEC protection for their zones. The steps in thisdocument are designed to ensure that all DNSKEY records and all DS[RFC4509] records (and the rest of a zone records) are properlyvalidatable by validating resolvers throughout the entire process.¶

Unfortunately, there are a number of these steps that are challengingto accomplish either because the timing is tricky to get right orbecause current software doesn't support automating the processeasily. Some examples:¶

1. The second step in Section 4.1.4 of[RFC6781] requires that a newkey with the new algorithm (which we refer to as K_new) be created,but not yet published. This step also requires that both the oldkey (K_old) and K_new sign and generate signatures for the zone,but with only the K_old key is published even though signaturesfrom K_new are included. After this odd mix has been published fora sufficient time length, based on the TTL, can K_new be safelyintroduced and published into the zone as well.¶
2. The new algorithm to be deployed isn't supported in the existingDNSSEC signing software and it is not possible (or not desired) tomove the private key into the DNSSEC signer that supports the newalgorithm choice.¶

Although many DNSSEC signing solutions may automate the algorithmrollover steps (making operator involvement unnecessary), many othertools do not support automated algorithm updates. In theseenvironments, the most challenging step is requiring that certainRRSIGs be published without the corresponding DNSKEYs that createdthem. This will likely require operators to use a text editor on thecontents of a signed zone to carefully select zone records to extractbefore publication. This introduces potentially significant operatorerror(s).¶

This document proposes an alternate, potentially more operationallyrobust but less secure, approach to performing algorithm DNSKEYrollovers for use in these situations.¶

2.Temporary transition mechanisms

3.Operational considerations

The process of replacing a DNSKEY with an older algorithm, such asRSAMD5 or RSASHA1 with a more modern one such as RSASHA512 orECDSAP256SHA256 can be a daunting task if the zone's current toolingdoesn't provide an easy-to-use solution. This is the case for zoneowners that potentially use command line tools that are integratedinto their zone production environment.¶

This document describes an alternative approach to rolling DNSKEYalgorithms that may be significantly less prone to operationalmistakes. However, understanding of the security considerations ofusing this approach is paramount.¶

The document recommends waiting 2 times TTL values in certain casesfor added assurance that the waiting period is long enough for cachesto expire. In reality, waiting only 1 TTL may be sufficient assumingall clocks around the world are operating with perfection.¶

4.Security considerations

DNSSEC provides an data integrity protection for DNS data. Thisdocument specifically calls out a reason why a zone owner may desireto deliberately turn off DNSSEC while changing the zone's DNSKEY'scryptographic algorithms. Thus, this is deliberately turning offsecurity which is potentially harmful if an attacker knows when thiswill occur and can use that time window to launch DNS modificationattacks (for example, cache poisoning attacks) against validatingresolvers or other validating DNS infrastructure.¶

Most importantly, this will deliberately break certain types of DNSrecords that must be validatable for them to be effective. Thisincludes for example, but not limited to, all DS records for childzones, DANE[RFC6698][RFC7671][RFC7672], PGP keys[RFC7929],and SSHFP[RFC4255]. Zone owners must carefully consider whichrecords within their zone depend on DNSSEC being available beforeusing the procedure outlined in this document.¶

Given all of this, it leaves the question of: "why would a zone ownerwant to deliberately turn off security temporarily then?", to whichthere is one principal answer. Simply put, if the the complexity ofdoing it the correct way is difficult with existing tooling then thechances of performing the more complex procedure and introducing anerror, likely making the entire zone unavailable during that timeperiod, may be significantly higher than the chances of the zone beingattacked during the transition period of the simpler approach wherezone availability is less likely to be impacted. Simply put, aninvalid zone created by a botched algorithm roll is potentially worsethan an unsigned but still available zone.¶

Appendix A.Acknowledgments

The author has discussed the pros and cons of this approach withmultiple people, including:¶

• Viktor Dukhovni¶
• Warren Kumari.¶
• Tuomo Soini¶
• Paul Wouters¶

Appendix B.Github Version of this document

While this document is under development, it can be viewed, tracked,issued, pushed with PRs, ... here:¶

https://github.com/hardaker/draft-hardaker-dnsop-intentionally-temporarily-insecure¶

Author's Address