Versions Compared

Old Version 167

changes.mady.by.user Kolawole Adebisi-Adeolokun

Saved on

compared with

New Version 182

changes.mady.by.user Kolawole Adebisi-Adeolokun

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Follow principles/patterns of RESTCONF RFC-8040 https://datatracker.ietf.org/doc/html/rfc8040
Follow principles/patterns of yang-patch RFC-8072 https://datatracker.ietf.org/doc/html/rfc8040
Follow principles/patterns of RESTCONF NMDA RFC-8527 https://datatracker.ietf.org/doc/html/rfc8527

Requirements

Note

Please note this section was added long after the implementation and focuses on characteristic only.

Issues & Decisions


IssuesNotesDecisions
1KPI for De-registration of 100 CM-handlesThis was mentioned. Was this ever agreed, is this a valid use case that needs to be covered together with the Registration ?

Not priority for now, but acceptable if we match the registration req.
#2 for de-reg  
kieran mccarthy Kolawole Adebisi-Adeolokun 

2DMI delayCould we get some feedback on DMI-delays for other use cases as not mentioned in FS document
Awaiing

Awaiting for ETH feedback AP On Kolawole Adebisi-Adeolokun and Csaba Kocsis

Characteristics - WIP

It is proposed that reported characteristics will be used as a baseline for NCMP when agreed and sign-off.


Provided

3

Number of instances


In some cases, ETH have used 2 instances, can we verify the number of instances for each use case.

Some of the req were defined per instance and resources used : Identify which of these ? 

Agreed to;

CPS use 1 instance currently, but should focus on aligning performance with 2 instances for all use case 

Kolawole Adebisi-Adeolokun Csaba Kocsis  

4Input Load Distribution the CM-handle search and ID search

Currently has 5 parallel request between them distributed at 2.5 each. This fractional distribution isn't feasible for parallel processing; the load should be allocated as whole numbers. Load needs to be distributed at. Would it be acceptable to adjust this distribution to either 2 or 3 parallel requests each (and vice versa ) without any negative repercussions?


Agreed to do 6 parallel request combined total and divide the load to 3 parallel request each

Kolawole Adebisi-Adeolokun Csaba Kocsis  

5Regarding CM-handle search and ID search

FS only identified Module performance, are there any testing done towards a combined search of properties and modules in a single query


Confirmed no other testing was previously done on this.....  

CPS have the capabilities to do mixed testing. ETH tbc on if they want to consider this ( Csaba Kocsis )


Requirements

Note

Please note this section was added long after the implementation and focuses on characteristic and enhancements after this study only.

Excerpt Include
DW:CPS-2249: NCMP To support Conflict Handling (Policy Execution) using External Service
DW:CPS-2249: NCMP To support Conflict Handling (Policy Execution) using External Service

Characteristics

It is proposed that reported characteristics will be used as a baseline for NCMP when agreed and sign-off.


Operation
OperationTotal CM-handles registered
Concurrent requests/parallelDMI Delay
Rest response
Response sizePerformance Requirement
(Blue Stone tablet KPI)		NotesSign-Off
1Registration of
100
20,000 CM-handles
20k Nodes (200 operations
(in batches of 100)1 (requests are sequential)

100 ms to get module references

1000

1,000 ms to get module resources

N/A

  • 11
NEs
  • CM-Handles/second as per Stone Tablet
  • NCMP should target 22 NEs/Second 

    • Currently without Modulesettag, when // start using Modulesettag, we'll need to revisit these numbers. 

    • Currently as per FS -
    • module references - 20k 
    • get module resources -1000 unique module ref. 
    • 5 different types of Nodes
    • E2E Wich include module conversion warm-up
    • NCMP Budget: 22 Cm Handles/second 
    1. Batch Size: 100 (per request)
      Not using Module Set Tags
    2. Time measured start of first rest-call until all cm handle states READY
    3. 1,000 unique module references.
    4. Five different types of Nodes. So 5 requests for Module Resources. Avg 200 modules each.
    ST 

    Csaba Kocsis Toine Siebelink 

    2De-registration of 100 CM-handles
    20k Nodes (200 operations)2.5 parallel (5)
    1 (requests are sequential)

    No Module delays

    N/A

    • 11 NEs/second
    • NCMP should target 22 NEs/Second 

    De-registration is currently not mentioned in Stone Tablet KPI or FS, however we have agreed to match the performance of registration for now as de-reg is also not a priority at this point in time




     

    Csaba Kocsis 

    Toine Siebelink Kolawole Adebisi-Adeolokun 

    3CM-handle ID search with Module filter
    20k Nodes

    3
    Run in parallel with #4

    		N/A
    100byte*20k = 

    TBD will be derived when testing is done in ETH envi Daniel Hanrahan 

    Combine test3&4 to = 5 (2.5 each)
    20,000 CM Handles i.e.
    100*20.000 = 2MB

    As provided byCsaba Kocsis 0.625 seconds/Operation

    		FS stated 5 parallel request for both ID search and search.  CPS to run with 3 parallel each for both ID search and Search meaning a combine total of 6 parallel request 

    Kolawole Adebisi-Adeolokun Csaba Kocsis 

    TDB

    4CM-handle search with Module filter
    20k Nodes2.5 parallelN/A500bytes*20k per handle = TBD will be derived when testing is done in ETH envi Daniel Hanrahan TDBAwaiting input from eth
    3
    Run in parallel with #3    		N/A20,000 CM Handles i.e.
    500*20.000 = 10MB

    As provided by Csaba Kocsis 13 seconds/Operation

    		FS stated 5 parallel request for both ID search and search.  CPS to test with 3 parallel each for both ID search and Search meaning a combine total of 6 parallel request 

    Kolawole Adebisi-Adeolokun Csaba Kocsis 

    5Synchronous single CM-handle pass-through read
    20k Nodes

    10

    10
    Run in parallel with #6

    300 ms

    Csaba Kocsis

    5 KB

    25

    op

    (parallel) request/sec

    shall not exceed

    Read are done in parallel with

    Writes

    Write

    Kolawole Adebisi-Adeolokun Csaba Kocsis  

    TDBAwaiting input from eth

    6Synchronous single CM-handle pass-through write (CUD)
    20k Nodes

    10

    10
    Run in parallel with #5

    670 ms

    Csaba Kocsis 

    5kb

    5 KB

    13

    ops/sec 

    (parallel) request/sec

    		No response is expected
    TDB

    Kolawole Adebisi-Adeolokun Csaba Kocsis 

    7Batch/Bulk Read

    1 read request with  200 cmHandle per second



    150 cmhandles/sec



    Notes

    1. This is for mixed TCs
    2. Single KPIs will be monitored in NCMP owned pipeline with our performance every day(2 hrs interval) - Performance

    ...