Last Call Review of draft-ietf-tram-turn-server-discovery-07
review-ietf-tram-turn-server-discovery-07-genart-lc-droms-2016-08-10-00

I am the assigned Gen-ART reviewer for this draft. The General Area
Review Team (Gen-ART) reviews all IETF documents being processed by
the IESG for the IETF Chair. Please resolve these comments along with
any other Last Call comments you may receive.

For more information, please see the FAQ at

<

http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.

Document: draft-ietf-tram-turn-server-discovery-08
Reviewer: Ralph Droms
Review Date: 2016-08-09
IETF LC End Date: 2016-08-11
IESG Telechat date: unknown

Summary:

This draft is on the right track but has open issues, described in the
review.

The draft is well-written and appears to be ready for publication,
except as noted below.

Major issues:

Section 5, DNS Service Discovery, includes more details about DNS
Service Discovery (DNS-SD) than is necessary for this specification.
While it can be useful to repeat some specific details of another
specification for, there is a danger in writing too many details that
may not be entirely in agreement with the published specification.  In
the case of this document, I suggest that section 5 be rewritten to
just refer to DNS Service discovery, with a minimum of explanation.
The example is useful ... although I think some of the details in the
example ought to be changed.  The use of DNS-SD over unicast DNS and
multicast DNS can be mentioned in a sentence somewhere in section 5,
as the use of DNS-SD is otherwise identical.  I would leave out
section 5.1 altogether.

Looking at the IANA "Service Name and Transport Protocol Port Number
Registry"
<www.iana.org/assignments/service-names-port-numbers/service-names-port-numbers.xhtml>,
I see that TURN is registered as using service name "turn", rather
than "turnserver" as in the example.  Also in the example, the
instance name "example.com" might be problematic, as the instance is
usually just a single label.  In fact, I interpret the text in the
document to describe the instance name as a single label.  It might be
worth experimenting to see how DNS-SD libraries deal with a label like
"example.com", or perhaps simply change instance in the example to
something like "exampleco TURN Server"

Minor issues:

Section 5 mentions the use of a TXT record to carry additional information
about the TURN service instance.  Are there any conventions for the name/value
pairs carried in the TXT record?  If not, I think there should be a note that
any name/value pairs in the TXT record are left to local definition.

Editorial issues:

I suggest using the example.com domain rather than local in the example for
clarity.  Perhaps also change the intro sentence for the example:

OLD:
 For example, TURN server advertises the following DNS records :
NEW:
 For example, the following DNS records would be used for a TURN server with
 instance name "exampleco TURN Server" providing TURN service over UDP on port
 5030:

It would help readability if the columns in the DNS records in the example
could be lined up; something like (apologies if your mail reader changes the
column alignments and if I don't have the quoting right):

_turnserver._udp.local.
PTR     "exampleco TURN Server"._turn._udp.local.

"exampleco TURN Server"._turn._udp.local.
SRV     0 0 5030 example-turn-server.local.

example-turn-server.local.
A       198.51.100.2

example-turn-server.local.
AAAA    2001:db8:8:4::2

Similarly, it would help readability if the list of DNS records for S-NAPTR
resolution were formatted in aligned columns.

In section 3, does "on top of" mean "in addition to" or "instead of"?

Attachment:

signature.asc

Description:

 Message signed with OpenPGP using GPGMail