Last Call Review of draft-ietf-dnsop-root-loopback-02
review-ietf-dnsop-root-loopback-02-opsdir-lc-romascanu-2015-08-13-00

I have reviewed this document as part of the Operational directorate's

ongoing effort to review all IETF documents being processed by the IESG.  These

comments were written with the intent of improving the operational aspects of
the

IETF drafts. Comments that are not addressed in last call may be included in AD
reviews

during the IESG review.  Document editors and WG chairs should treat these
comments

just like any other last call comments.



This I-D aims Informational status. It describes an optimization of the DNS
recursive resolvers to speed the queries and hide them from external
observation by creating an up-to-date root zone server on an IPv4 loopback
address on the same
 host as the recursive server.



If implemented the method described by this I-D has an operational impact. I
have used the Operational Considerations checklist in Appendix A1 of RFC 5706
for the review result:



1.  Has deployment been discussed?  See Section 2.1.



       *  Does the document include a description of how this protocol

          or technology is going to be deployed and managed?



       *  Is the proposed specification deployable?  If not, how could

          it be improved?



       *  Does the solution scale well from the operational and

          management perspective?  Does the proposed approach have any

          scaling issues that could affect usability for large-scale

          operation?



       *  Are there any coexistence issues?



Yes – deployment is discussed in section 1. Because of the operational risks
discussed in the document distributions of recursive DNS servers MUST NOT
include configuration for the design defined in this document.



   2.  Has installation and initial setup been discussed?  See

       Section 2.2.



       *  Is the solution sufficiently configurable?



       *  Are configuration parameters clearly identified?



       *  Are configuration parameters normalized?



       *  Does each configuration parameter have a reasonable default

          value?



       *  Will configuration be pushed to a device by a configuration

          manager, or pulled by a device from a configuration server?



       *  How will the devices and managers find and authenticate each

          other?



Yes – to the extent that configuration means here keeping an up-to-date copy of
the root zone server



   3.  Has the migration path been discussed?  See Section 2.3.



       *  Are there any backward compatibility issues?



N/A – this is not a new version of the protocol, but an optional incremental
optimization



   4.  Have the Requirements on other protocols and functional

       components been discussed?  See Section 2.4.



       *  What protocol operations are expected to be performed relative

          to the new protocol or technology, and what protocols and data

          models are expected to be in place or recommended to ensure

          for interoperable management?



N/A – there is no interoperability issue as long as the root zone server is an
up-to-date replica of the external server.



   5.  Has the impact on network operation been discussed?  See

       Section 2.5.



       *  Will the new protocol significantly increase traffic load on

          existing networks?



       *  Will the proposed management for the new protocol

          significantly increase traffic load on existing networks?



       *  How will the new protocol impact the behavior of other

          protocols in the network?  Will it impact performance (e.g.,

          jitter) of certain types of applications running in the same

          network?



       *  Does the new protocol need supporting services (e.g., DNS or

          Authentication, Authorization, and Accounting - AAA) added to

          an existing network?



Yes – Section 1 makes clear that the deployment of this method leads to faster
negative responses, and probably has little effect on getting faster positive
responses.



   6.  Have suggestions for verifying correct operation been discussed?

       See Section 2.6.



       *  How can one test end-to-end connectivity and throughput?



       *  Which metrics are of interest?



       *  Will testing have an impact on the protocol or the network?



Yes – Section 4 includes recommendations about testing proper operation.



   7.  Has management interoperability been discussed?  See Section 3.1.



       *  Is a standard protocol needed for interoperable management?



       *  Is a standard information or data model needed to make

          properties comparable across devices from different vendors?



N/A



   8.  Are there fault or threshold conditions that should be reported?

       See Section 3.3.



       *  Does specific management information have time utility?



       *  Should the information be reported by notifications?  Polling?

          Event-driven polling?



       *  Is notification throttling discussed?



       *  Is there support for saving state that could be used for root

          cause analysis?



N/A



   9.  Is configuration discussed?  See Section 3.4.



       *  Are configuration defaults and default modes of operation

          considered?



       *  Is there discussion of what information should be preserved

          across reboots of the device or the management system?  Can

          devices realistically preserve this information through hard

          reboots where physical configuration might change (e.g., cards

          might be swapped while a chassis is powered down)?



Yes – Section 1 recommends that

distributions of recursive DNS servers MUST NOT include configuration for the
design defined in this document.



A few more comments.



1.



The Abstract mentions ‘the cost of adding some operational fragility to the
operator’. I am scratching my head about what this means. Adding one more
component may always add some ‘fragility’ but in this
 case the result is a behavior optimization, so is this really a cost to be
 concerned about?

2.



Recursive resolver software such as BIND, Unbound, NSD are mentioned –
references may help

3.



SOA is never expanded

4.



Security Considerations may mention the positive effect of keeping the queries
internal, thus preventing observation of traffic on the network between the
recursive resolver and one or more of the DNS roots.



I hope this helps.



Dan