NCMP: Read data operation resource API Performance for multiple cm-handles

Read data operation resource for multiple cm-handles

Capabilities

Test Configurations

#

Environment/Workload

Description

#

Environment/Workload

Description

1

Teed on 

laptop :
Processor :          Intel© Core™ i7-8750H CPU @ 2.20GHz × 6
Installed RAM :   32.0 GB
Edition :               Linux Mint 21.1

2

Shell script, postman request and generated txt file with 20,000 cm handle ids

3

Number of CPS Instance

1

4

Number Modules per Cm Handle

10

5

Total number of cm handles

20,000

6

Batch-size

200 cm handles per request

7

Response payload size

~2 KB per cm handles

8

Concurrent request

12

9

Number of DMI PLugins

1

10

Commit ID 

9830d1aec56da5cec195f6d15c318b58f642c9b7

11

Commit ID link

https://gerrit.onap.org/r/gitweb?p=cps.git;a=commit;h=9830d1aec56da5cec195f6d15c318b58f642c9b7



#

Component

Configuration Parameters

Value

#

Component

Configuration Parameters

Value

1

CPS

notifactions:enabled

true

2

CPS

config.additional.notification.data-updated.filters.dataspaces.enabled-dataspaces

"NON-EXISTING-DATASPACE"

3

NCMP

modules-sync-watchdog:async-executor:parallelism-level

10

Summary of test results

Description

Expectation

Result

Description

Expectation

Result

Concurrency (parallel thread requests)

12

12

Test Start (received first kafka response message)

N/A

14:07:50

Test End (received last message)

< 14:10:10

14:08:03

Total time taken to publish 20,000 cm handle responses to client topic

< 200 sec

13.2 sec

Average Time per batch (of 200 cm handles)

< 2 sec

0.132 sec

Pace of cm handle response to client topic ()

100 cm handles/ sec

1,514 cm handles/ sec

Notes

  1. Above test has been executed with just (stubbed) 1 DMI Plugin. 

  2. More DMI Plugin should make it even faster

  3. As per usual NCMP performance is measured without delays in the Plugin Layer