Re: [dnssd] DNS Name Autoconfiguration for Home Network Devices

["Mr. Jaehoon Paul Jeong" <[e-mail]>]

Hi Robby,
Thanks for your questions.

>Hi Paul,
>
>If the actual DNS name is used, I would be concerned by (at least) the
following:
>- Parsing — clearly there will be non-compliant DNS names out there, that
these devices will have to parse and trash.  Looking at your draft, how
would they know?

In my proposal, an IoT device does not process the DNS query because it
does not run mDNS
for the DNS name resolution service.
Since the DNS name of the IoT device is registered into a DNS server via a
router that
is located at the same subnet with the IoT device.

>- Extensibility — in the future how could new information be included?
New categories?

When a new category is added into a new device's repository as factory
default,
such a new category will be used by the device for generating its DNS name.

>- Size — as the target is IoT, bytes on the "wire" matter and this
descriptive information is going to be quite large in the DNS format

In the case where the DNS name of an IoT device cannot be accommodated and
delivered by a MAC frame
to a router for the DNS name registration, multiple MAC frames having parts
of the DNS name will be sent and
merged at a gateway or router within the communication range of the IoT
device.
For 6LoWPAN, we can consider a compact DNS name format for IoT device for
the given device category
and model.

Paul

>
>My preference would be for devices to support some resource that is
discovered via DNS-SD — see my other email for details.
>
>- Robby

===========================
Mr. Jaehoon (Paul) Jeong, Ph.D.
Assistant Professor
Department of Software /
Department of Computer Engineering
Sungkyunkwan University
Office: [phone]
Mobile: [phone]
Fax: [phone]
Email: [e-mail], [e-mail]
CPS Lab Website: http://cpslab.skku.edu
Personal Homepage: http://cpslab.skku.edu/people-jaehoon-jeong.php

On Fri, Nov 21, 2014 at 12:57 AM, Simpson, Robby (GE Energy Management) <
[e-mail]> wrote:

> Hi Paul,
>
> If the actual DNS name is used, I would be concerned by (at least) the
> following:
> - Parsing — clearly there will be non-compliant DNS names out there, that
> these devices will have to parse and trash.  Looking at your draft, how
> would they know?
> - Extensibility — in the future how could new information be included?
> New categories?
> - Size — as the target is IoT, bytes on the "wire" matter and this
> descriptive information is going to be quite large in the DNS format
>
> My preference would be for devices to support some resource that is
> discovered via DNS-SD — see my other email for details.
>
> - Robby
>
> From: "Mr. Jaehoon Paul Jeong" <[e-mail]<mailto:
> [e-mail]>>
> Date: Monday, November 17, 2014 at 8:29 PM
> To: Charles Simpson <[e-mail]<mailto:[e-mail]>>
> Cc: Dave Thaler <[e-mail]<mailto:[e-mail]>>,
> Brian Haberman <[e-mail]<mailto:[e-mail]>>,
> Myung-Ki Shin <[e-mail]<mailto:[e-mail]>>, "
> [e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:
> [e-mail]>>, Jung-Soo Park <[e-mail]<mailto:[e-mail]>>,
> Ted Lemon <[e-mail]<mailto:[e-mail]>>, Sejun
> Lee <[e-mail]<mailto:[e-mail]>>
> Subject: Re: [dnssd] DNS Name Autoconfiguration for Home Network Devices
>
> Hi Robby,
> Yes, my intent is to to include semantics (e.g., device category, device
> vendor, and device model)
> in a standardized fashion into the DNS names of low-capacity IoT devices.
>
> I intend to place machine-interpretable semantics into the actual DNS
> names themselves
> so that machines can use the DNS name for the identification of other
> machines
> without additional service discovery through mDNS.
>
> The localization of devices is also possible with the DNS name
> because the DNS name can include the location of the device,
> as describe in Section 6 (Location-Aware DNS Name Configuration) in my
> draft:
> http://datatracker.ietf.org/doc/draft-jeong-homenet-device-name-autoconf/
>
> Thanks for your good opinion.
>
> Paul
>
> ===========================
> Mr. Jaehoon (Paul) Jeong, Ph.D.
> Assistant Professor
> Department of Software /
> Department of Computer Engineering
> Sungkyunkwan University
> Office: [phone]
> Mobile: [phone]
> Fax: [phone]
> Email: [e-mail]<mailto:[e-mail]>,
> [e-mail]<mailto:[e-mail]>
> CPS Lab Website: http://cpslab.skku.edu
> Personal Homepage: http://cpslab.skku.edu/people-jaehoon-jeong.php
>
> On Tue, Nov 18, 2014 at 6:15 AM, Simpson, Robby (GE Energy Management) <
> [e-mail]<mailto:[e-mail]>> wrote:
> It seems to me that part of your intent is to include semantics (e.g.,
> device category, device vendor, device model) in a standardized fashion
> into the DNS name.
>
> On the other hand, while we often apply semantics to DNS names currently
> for human readers, these semantics typically are not standardized for
> machines.  For that, we have DNS-SD.
>
> As an example from the IoT space, we use both mDNS and DNS-SD for SEP 2.0
> (IEEE 2030.5).  While the DNS names often reflect aspects such as device
> manufacturer and category, these are not meant to be machine interpretable
> in SEP 2.0.  Rather, we use DNS-SD to advertise various functionality that
> is machine interpretable.
>
> Perhaps I am misinterpreting, but is your intent to place
> machine-interpretable semantics into the actual DNS names themselves?
>
> Thanks,
> Robby
>
>
> Robby Simpson, PhD
>
> System Architect
>
> GE
>
> Digital Energy
>
> M: [phone]
>
> [e-mail]<mailto:[e-mail]>
>
>
> From: "Mr. Jaehoon Paul Jeong" <[e-mail]<mailto:
> [e-mail]><mailto:[e-mail]<mailto:
> [e-mail]>>>
> Date: Saturday, November 15, 2014 at 1:22 AM
> To: Dave Thaler <[e-mail]<mailto:[e-mail]
> ><mailto:[e-mail]<mailto:[e-mail]>>>
> Cc: Brian Haberman <[e-mail]<mailto:
> [e-mail]><mailto:[e-mail]<mailto:
> [e-mail]>>>, Myung-Ki Shin <[e-mail]<mailto:
> [e-mail]><mailto:[e-mail]<mailto:[e-mail]>>>, "
> [e-mail]<mailto:[e-mail]><mailto:[e-mail]<mailto:
> [e-mail]>>" <[e-mail]<mailto:[e-mail]><mailto:
> [e-mail]<mailto:[e-mail]>>>, Jung-Soo Park <[e-mail]
> <mailto:[e-mail]><mailto:[e-mail]<mailto:[e-mail]>>>,
> Ted Lemon <[e-mail]<mailto:[e-mail]><mailto:
> [e-mail]<mailto:[e-mail]>>>, Sejun Lee <
> [e-mail]<mailto:[e-mail]><mailto:
> [e-mail]<mailto:[e-mail]>>>
> Subject: Re: [dnssd] DNS Name Autoconfiguration for Home Network Devices
>
> Dave,
> Thanks for your clarification.
>
> In Page 32 in RFC 6762, there is the recommended course of action after
> probing and failing, but
> there is no text about a random ID selection.
> Anyway, we can perform a random ID selection for the uniqueness of a DNS
> name, but
> the readability for such a DNS name is not good for the users.
>
> My original intention for DNS name generation is to include device
> category (e.g., refrigerator),
> device vendor (e.g., Samsung), device model (e.g., RH269LP).
> This name itself delivers much information to users and mobile  smart
> devices (e.g., smartphone or smart TV)
> to represent the device icon visually.
>
> I am not sure this is enough answer for your last question.
> If you have more comments, please let me know.
>
> Paul
>
> ===========================
> Mr. Jaehoon (Paul) Jeong, Ph.D.
> Assistant Professor
> Department of Software /
> Department of Computer Engineering
> Sungkyunkwan University
> Office: [phone]
> Mobile: [phone]
> Fax: [phone]
> Email: [e-mail]<mailto:[e-mail]><mailto:
> [e-mail]<mailto:[e-mail]>>, [e-mail]
> <mailto:[e-mail]><mailto:[e-mail]<mailto:
> [e-mail]>>
> CPS Lab Website: http://cpslab.skku.edu
> Personal Homepage: http://cpslab.skku.edu/people-jaehoon-jeong.php
>
> On Fri, Nov 14, 2014 at 11:59 AM, Dave Thaler <[e-mail]
> <mailto:[e-mail]><mailto:[e-mail]<mailto:
> [e-mail]>>> wrote:
> Paul wrote:
> > For the regeneration and verification of a unique DNS name under DNS
> name conflict,
> > the solution in RFC 6762 recommends to use an incremental digit (such as
> 2, 3, 4, etc.)
> > by trial and error. In an IoT scenario where there will be many IoT
> devices of the same
> > type, such as light bulb in home or hotel here, this incremental
> numbering approach
> > will be costly and slow to let each IoT device have a unique DNS name,
> ...
>
> My reading is that RFC 6762 does not _require_ an incremental digit.  You
> can put in
> a random ID or MAC-derived ID or something else highly unlikely to collide.
> As such, it should not be "costly and slow".  Indeed RFC 6762 does not
> specify what
> you have to do.   Would it be possible to recast your draft as
> "how to choose a unique ID and use RFC 6762" ?
>
> -Dave
>
>
>