Re: [6tsch] work item 2: centralized route and track computation with PCE

["Pascal Thubert (pthubert)" <[e-mail]>]

My suggestion, Thomas,

is to keep that text till the BoF meeting to help explain what we have to do, and probably to remove it after that.

Because our charter will appear too thick...

Pascal

From: [e-mail] [mailto:[e-mail]] On Behalf Of Thomas Watteyne
Sent: mardi 23 juillet 2013 17:32
To: [e-mail]
Subject: Re: [6tsch] work item 2: centralized route and track computation with PCE

Pascal, all,

I'm starting from the text currently in the repo (https://bitbucket.org/6tsch/charter-ietf-6tsch/src/master/charter-ietf-6tsch-00.txt):

2. Produce requirements and an overall architecture for "6TSCH centralized routes and tracks management" describing the mechanism by which an external Path Computation Element (PCE) can communicate with devices in a 6TSCH network in order to maintain the routing, forwarding and scheduling states.
In detail, the WG will consider operations:
(1) from the device routing layer to the PCE for
        (i) sending topological information and routing metrics, and
        (ii) getting on-demand routing information; and
(2) from the device 6top sublayer to the PCE for
        (a) exporting neighboring and link quality statistics and
        (b) requesting to (re)compute and install a track that meets explicit flow requirements.
The requirements and architecture documents would aim at informational status and as a result standard track work may be triggered within this group or other groups such as the PCE WG.

I believe we have a consensus on the green parts. The issue, as highlighted by Michael, is that the blue parts may read like we are trying to solve the problem in the charter.

We have 3 options:
1. leave as is (or minor rewording)
2. drop the blue part
3. simplify to 1-2 sentences

The charter is too long already, and work item 2 is now by far the largest one. I would therefore vote for option 2 (dropping). IMO, the statement "maintain the routing, forwarding and scheduling state" sums it up.

What is missing is the fact that the nodes need to inform the PCE of their topological and traffic requirements. We could therefore change this work item:

2. Produce requirements and an overall architecture for "6TSCH centralized routes and tracks management" describing the mechanism by which an external Path Computation Element (PCE) can communicate with devices in a 6TSCH network in order to retrieve topological information and traffic requirements, and maintain the routing, forwarding and scheduling states.
The requirements and architecture documents would aim at informational status and as a result standard track work may be triggered within this group or other groups such as the PCE WG.

Thoughts?
Thomas


On Tue, Jul 23, 2013 at 4:51 PM, Pascal Thubert (pthubert) <[e-mail]<mailto:[e-mail]>> wrote:
Hello Raghu:

A couple of things -
1) In detail, the WG will consider operations (1) by the device routing layer for
In my understanding, this interaction will be between the PCE and the networking layer. Does this fall under the general 6top scope to even 'define' the requirements? I understand that it may be necessary for operational. IMO, it is better if it is defined by PCE. This part could be an informational document though.
[] This is why the text says "consider" instead of "define". Also the beginning of the sentence changed to say "requirements and an overall architecture".
Also it was lost in the discussions but the trailer of that section now says:
"The requirements and architecture documents would aim at Informational status and as a result standard track work may be triggered within this group or other groups such as the PCE WG."
I uploaded this to the repo but we can certainly still amend it.

(2) by the device 6top sublayer for exporting neighboring and link quality statistics to the PCE
We can drop the device. Instead - by the 6top sublayer for exporting neighboring and link quality statistics to the PCE. Same comment for (1). 'by the routing layer'
[] I do no mind either way. What do others think?

I second the addition of "scheduling states" to the text as discussed below.
[] Great; we are converging. Maybe we can move the discussion toward sthe adoption (or not) of the security as a work item? Since you are the one who introduced the question, maybe you can start the thread?

Cheers,

-raghuram

From: Maria Rita PALATTELLA <[e-mail]<mailto:[e-mail]>>
Date: Tuesday, July 23, 2013 6:39 AM
To: "Pascal Thubert (pthubert)" <[e-mail]<mailto:[e-mail]>>, raghuram sudhaakar <[e-mail]<mailto:[e-mail]>>, Thomas Watteyne <[e-mail]<mailto:[e-mail]>>
Cc: "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: RE: work item 2: centralized route and track computation with PCE

Pascal,
Thanks for your explanation. I agree with you.
About the definition of the work item 2, I would keep the last part as it was before, i.e.,

2. Produce requirements and an overall architecture for  "6TSCH centralized routes and tracks management"
describing the mechanism by which an external Path Computation Element (PCE) can communicate with
devices in a 6TSCH network in order to maintain the routing, forwarding and scheduling states.

In detail, the WG will consider operations (1) from  the device routing layer to the PCE for
(i) sending topological information and routing metrics, and (ii) getting on-demand routing information;
and (2) from the device 6top sublayer  to the PCE for (a) exporting neighboring and link quality statistics
and (b) requesting to (re)compute and install a track that meets explicit flow requirements."


Maria Rita



From: Pascal Thubert (pthubert) [mailto:[e-mail]]
Sent: Tuesday, July 23, 2013 3:15 PM
To: Maria Rita PALATTELLA; Raghuram Sudhaakar (rsudhaak); Thomas Watteyne
Cc: [e-mail]<mailto:[e-mail]>
Subject: work item 2: centralized route and track computation with PCE

Hello Maria Rita.

In one hand, I agree that the device only receives a schedule information, though it is not all of the device schedule.
At the same time, a device A instructs the PCE to compute an end-to-end track between A and Z and once done, the PCE must come back with the individual schedules of all the nodes on the way. We are not clear yet as of how the individual schedules will be pushed to intermediate nodes but that must happen and may reuse PCEP.

"
2. Produce requirements and an overall architecture for  "6TSCH centralized routes and tracks management"
describing the mechanism by which an external Path Computation Element (PCE) can communicate with
devices in a 6TSCH network in order to maintain the routing, forwarding and scheduling states.

In detail, the WG will consider operations (1) from  the device routing layer to the PCE for
(i) sending topological information and routing metrics, and (ii) getting on-demand routing information;
and (2) from the device 6top sublayer  to the PCE for (a) exporting neighboring and link quality statistics
and (b) requesting to (re)install a schedule along a track that meets explicit flow requirements
between the track endpoints.
"

What do you think?


Pascal

From: Maria Rita PALATTELLA [mailto:[e-mail]]
Sent: mardi 23 juillet 2013 14:22
To: Pascal Thubert (pthubert); Raghuram Sudhaakar (rsudhaak); Thomas Watteyne
Cc: [e-mail]<mailto:[e-mail]>
Subject: RE: work item 2: centralized route and track computation with PCE

Pascal,
The new definition is for sure more complete and shows better what we are planning to do with the 6TSCH centralized approach.
Here are my few comments:

-          I would add a reference to the schedule. Because the PCE will take care of it, and not only of routing and forwarding states.

-          Also in the last part, I would talk in term of schedule, instead of track.

It reads as following:

2. Produce requirements and an overall architecture for  "6TSCH centralized routes and tracks management"
describing the mechanism by which an external Path Computation Element (PCE) can communicate with
devices in a 6TSCH network in order to maintain the schedule, and  routing/forwarding states.

In detail, the WG will consider operations (1) from  the device routing layer to the PCE for (i) sending topological information and routing metrics, and (ii) getting on-demand routing information;
and (2) from the device 6top sublayer  to the PCE for (a) exporting neighboring and link quality statistics and (b) requesting a (new) schedule that meets explicit flow requirements.

Maria Rita

From: Pascal Thubert (pthubert) [mailto:[e-mail]]
Sent: Tuesday, July 23, 2013 12:34 PM
To: Raghuram Sudhaakar (rsudhaak); Maria Rita PALATTELLA; Thomas Watteyne
Cc: [e-mail]<mailto:[e-mail]>
Subject: work item 2: centralized route and track computation with PCE

Hello Raghu and all:

About item 2 6TSCH Centralized Management:

Since we wrote the charter a few months ago we have been digging and it appears that we are probably not ready for going standard track right away with a 6TSCH document.
- In one hand, we understand better that the PCE is the component we need for both centralized route AND track computation.
- OTOH, we realized that PCEP only covers a limited portion of the required standard work and we realize that we need to dig deeper in what needs to be done, in particular to locate the PCE(s), export topology and metrics to the PCE, and then install routes and tracks from the PCE.  We also understand that some of that work belongs to the PCE working group though it is unclear exactly how much that is.

Right now the charter has this:

2. Produce "6TSCH centralized management" to define the mechanism by which an
external Path Computation Element can communicate with the 6top protocol layer
of the different nodes in the network (1) to modify their TSCH schedule and
(2) to gather link quality statistics and data flow requirements.  The WG will
initially look at PCEP and CoAP.
Depending on the applicability of PCEP, this document will be targeted to
proposed standard.

In the light of our studies, I'd suggest this:
2. Produce requirements and an overall architecture for  "6TSCH centralized routes and tracks management"
describing the mechanism by which an external Path Computation Element (PCE) can communicate with
devices in a 6TSCH network in order to maintain routing and forwarding states. In detail, the WG will
consider operations (1) by the device routing layer for (i) sending topological information and routing
metrics to the PCE, and (ii) getting on-demand routing information; and (2) by the device 6top sublayer
for exporting neighboring and link quality statistics to the PCE and (2) requesting to (re)compute and
install a track that meets explicit flow requirements.

What do you guys think?


Pascal

From:[e-mail]<mailto:[e-mail]> [mailto:[e-mail]] On Behalf Of Raghuram Sudhaakar (rsudhaak)
Sent: lundi 22 juillet 2013 07:25
To: [e-mail]<mailto:[e-mail]>
Subject: [6tsch] slides for work items

All,
Attached are the slides for the work items. Please provide your comments.

Specifically, please provide feedback on whether we must have the security framework as a work item.

-raghuram