[OPSAWG] YANG based Collection and Aggregation Framework//: New Version Notification for draft-opsawg-poweff-00.txt

2023-10-31 Thread Qin Wu
Hi, Jan, Marisol:
I have read the latest version of 
draft-lindblad-tlm-philatelist
 and feel it is very interesting draft since it can aggregate data from various 
different data sources using different telemetry technologies.
First I see many metadata related work in this space , some of them are 
ongoing, some of them have already expired, e.g.,

1.   out of band metadata collection  
https://datatracker.ietf.org/doc/draft-ietf-opsawg-collected-data-manifest/

2.   in band metadata collection 
https://datatracker.ietf.org/doc/html/draft-ietf-netmod-node-tags-09

3.   Semantic Metadata Annotation  
https://datatracker.ietf.org/doc/draft-netana-opsawg-nmrg-network-anomaly-semantics/

4.   cloud-native metrics and time series format  
https://datatracker.ietf.org/doc/html/draft-richih-opsawg-openmetrics-00

5.   self-describing format for metadata abstract  
https://www.ietf.org/archive/id/draft-gray-sampled-streaming-03.txt

Secondly. I think the metadata can be used to collect telemetry context 
information, telemetry data classification information,
In your draft, the metadata collected by YANG based collection and aggregation 
framework seems more related to measurement related attributes,
Such as measurement unit, mathematic operation such as sum, average, max, 
min,etc, I remember XPATH 1.0 do support some of mathematic computation 
operation, such as max, min and logical operation such as AND and OR,
This has been discussed in 
https://www.ietf.org/archive/id/draft-ietf-netmod-eca-policy-01.
I can imagine the network device has already support some of mathematic 
computation operation, therefore can report summary statistics data using 
telemetry interface.

3.Regarding measurement quantities, I think open metrics also can be used to 
aggregate metrics related data, express all system states as numerical values;
counts, current values, enumerations, and Boolean states, I am wondering how 
these two work are different, whether 
draft-lindblad-tlm-philatelist
 can better align with
open metrics regarding metric aggregation.
4. Section 3.3 said:
“

Each flow is associated with one or more inputs, one output and a

   series of processing operations.  Each input flow and output flow may

   have an pre-processing or post-processing operation applied to it

   separately.  Then all the input flows are combined using one or more

   aggregation operations.

”
It is very interesting to keep track of flow from the data source to 
destination group and traverse a set of aggregation points,
It seems to provide a good visibility to the flow path within the analytics 
platform, I am wondering what is the benefit or goal to keep track of these 
flow path?
Is this implementation specific?

-Qin
发件人: OPSAWG [mailto:opsawg-boun...@ietf.org] 代表 Marisol Palmero Amador 
(mpalmero)
发送时间: 2023年10月28日 1:07
收件人: opsawg@ietf.org
抄送: Gonzalo Salgueiro (gsalguei) ; Jan Lindblad (jlindbla) 
; Snezana Mitrovic (snmitrov) ; 
emile.step...@orange.com; Per Andersson (perander) ; Esther 
Roure Vila (erourevi) ; emile.step...@orange.com
主题: [OPSAWG] FW: New Version Notification for draft-opsawg-poweff-00.txt

Dear OPSA WG,

Earlier this week, we posted a new draft that introduces a data model for power 
and energy related metrics :
https://datatracker.ietf.org/doc/draft-opsawg-poweff/
The focus is mainly on runtime information provided by power sensors, but also 
an extension to other related metrics and given attributes that will complement 
the representation of the energy consumed by the network device, implemented in 
hardware or software, as well as by specific network components.
This is a first-version approach where we see still challenges based on 
implementation.
Note: Some of those challenges are covered on Jan`s draft: 
draft-lindblad-tlm-philatelist

Along with POWEFF draft, we’ve also updated the version of the Sustainability 
Insights draft:
https://datatracker.ietf.org/doc/draft-almprs-sustainability-insights/
where we introduced an updated architecture reference diagram, that provides a 
more structured view of the functional blocks that might be part of where those 
attributes and metrics might be produced, processed, visualized, etc. We also 
have reviewed and added Use Cases that such framework could drive.

We greatly appreciate your thoughts and comments.

Many thanks,
Marisol Palmero


From: internet-dra...@ietf.org 
mailto:internet-dra...@ietf.org>>
Date: Friday, 20 October 2023 at 17:45
To: Gonzalo Salgueiro (gsalguei) 
mailto:gsalg...@cisco.com>>, Jan Lindblad (jlindbla) 
mailto:jlind...@cisco.com>>, Marisol Palmero Amador 
(mpalmero) mailto:mpalm...@cisco.com>>, Snezana Mitrovic 
(snmitrov) mailto:snmit...@cisco.com>>
Subject: New Version Notification

Re: [OPSAWG] YANG based Collection and Aggregation Framework//: New Version Notification for draft-opsawg-poweff-00.txt

2023-11-01 Thread Jan Lindblad (jlindbla)
Qin,

Excellent feedback! Thanks for taking the time to read and compose detailed 
comments.

I'm very grateful for the reading list; now I know what to do on the train to 
Prague :-) I was unaware of most of these, so very timely pointers. I am aware, 
however, that there is plenty of prior work in this field, and much going on 
right now in IETF and at vendors. I hadn't seen anything that addressed the 
Time Series Database (TSDB) integration, though, so I thought I'd write down 
what I had been thinking. My hope is that this draft will merge with existing 
work in the area.

Your observation that this approach favors traceability of data flows all the 
way back to individual sensors is gratifying; that is very much the core idea. 
Too many times have I said "Nice graph; so are you reading PSU voltage and 
current sensors on the device? Is that the true power, or reactive/apparent 
power? How accurate is that reading? Is the local cooling cost (i.e. fans) 
included?" and usually (and I hear many others reporting the same experience) 
such questions cannot be answered the same day. The measurements are not 
transparent.

Let's continue the discussion in Prague.

Best Regards,
/Jan


On 1 Nov 2023, at 06:50, Qin Wu  wrote:

Hi, Jan, Marisol:
I have read the latest version of 
draft-lindblad-tlm-philatelist
 and feel it is very interesting draft since it can aggregate data from various 
different data sources using different telemetry technologies.
First I see many metadata related work in this space , some of them are 
ongoing, some of them have already expired, e.g.,
1.   out of band metadata collection  
https://datatracker.ietf.org/doc/draft-ietf-opsawg-collected-data-manifest/
2.   in band metadata collection 
https://datatracker.ietf.org/doc/html/draft-ietf-netmod-node-tags-09
3.   Semantic Metadata Annotation  
https://datatracker.ietf.org/doc/draft-netana-opsawg-nmrg-network-anomaly-semantics/
4.   cloud-native metrics and time series format  
https://datatracker.ietf.org/doc/html/draft-richih-opsawg-openmetrics-00
5.   self-describing format for metadata abstract  
https://www.ietf.org/archive/id/draft-gray-sampled-streaming-03.txt

Secondly. I think the metadata can be used to collect telemetry context 
information, telemetry data classification information,
In your draft, the metadata collected by YANG based collection and aggregation 
framework seems more related to measurement related attributes,
Such as measurement unit, mathematic operation such as sum, average, max, 
min,etc, I remember XPATH 1.0 do support some of mathematic computation 
operation, such as max, min and logical operation such as AND and OR,
This has been discussed in 
https://www.ietf.org/archive/id/draft-ietf-netmod-eca-policy-01.
I can imagine the network device has already support some of mathematic 
computation operation, therefore can report summary statistics data using 
telemetry interface.

3.Regarding measurement quantities, I think open metrics also can be used to 
aggregate metrics related data, express all system states as numerical values;
counts, current values, enumerations, and Boolean states, I am wondering how 
these two work are different, whether 
draft-lindblad-tlm-philatelist
 can better align with
open metrics regarding metric aggregation.
4. Section 3.3 said:
“

Each flow is associated with one or more inputs, one output and a

   series of processing operations.  Each input flow and output flow may

   have an pre-processing or post-processing operation applied to it

   separately.  Then all the input flows are combined using one or more

   aggregation operations.


”
It is very interesting to keep track of flow from the data source to 
destination group and traverse a set of aggregation points,
It seems to provide a good visibility to the flow path within the analytics 
platform, I am wondering what is the benefit or goal to keep track of these 
flow path?
Is this implementation specific?

-Qin
发件人: OPSAWG [mailto:opsawg-boun...@ietf.org] 代表 Marisol Palmero Amador 
(mpalmero)
发送时间: 2023年10月28日 1:07
收件人: opsawg@ietf.org
抄送: Gonzalo Salgueiro (gsalguei) 
mailto:gsalg...@cisco.com>>; Jan Lindblad (jlindbla) 
mailto:jlind...@cisco.com>>; Snezana Mitrovic (snmitrov) 
mailto:snmit...@cisco.com>>; 
emile.step...@orange.com; Per Andersson 
(perander) mailto:peran...@cisco.com>>; Esther Roure Vila 
(erourevi) mailto:erour...@cisco.com>>; 
emile.step...@orange.com
主题: [OPSAWG] FW: New Version Notification for draft-opsawg-poweff-00.txt

Dear OPSA WG,

Earlier this week, we posted a new draft that introduces a data model for power 
and energy related metrics :
https://datatracker.ietf.org/doc/draft-opsawg-poweff/
The focus is mainly on runtime information provide