O

verview

• This document contains the results of Central Ordering testing scenarios to create order with PO lines plus Change instance connection and Receive all pieces for Ramsons [ECS] environments. 

Summary

...

In Test 3, the Create Order operation involved creating orders with 50 Order Lines, compared to 1 Order Lines in Test 1 and 10 Order Lines in Test 2. This explains the significant increase in response times for the Open Order and Unopen Order operations:

• Open Order: Response times increased nearly 10x in Test 3 compared to Test 1 due to the higher load (50 items per order).
• Unopen Order: Similarly, response times grew approximately 4-5x in Test 3.

For other operations (Create Order, Create Order Line, Receive Pieces, and Change Instance Connection), response times increased moderately but remained within acceptable ranges as the load increased.

Conclusion: The increase in item count per order directly impacted the performance of Open Order and Unopen Order, highlighting their sensitivity to load.

...

Overview

• This document contains the results of Acquisition Audit Log Performance testing for Ramsons [ECS] environments. 

Summary

• All tests were successfully completed, and all API calls executed without errors.
• The response time variation for different data sizes or changes in each Order, Order Line, or Piece history is around 5%.
• The duration for inserting one row remains consistent across different data sizes.
• All test were executed without any load on background for our system.

Test Runs 

Test Results

Test №1

PO with PO Lines (1) + locations (5) + Change instance connection (Move holdings)The following table collect transactions results for tests №1,  №2 and №3.

The following table collect API results for tests №1,  №2 and №3:

• GET audit-data/acquisition/order/order_id
• GET audit-data/acquisition/order-line/order_line_id
• GET audit-data/acquisition/piece/piece_id/status-change-history

The following table collect INSERT INTO duration results for different data size.

Comparison

Test №1-№2-№3

• The following table

...

• compares the average response times for all transactions between tests №1,  №2 and №3. 

• Open Order: Response times increased nearly 10x in Test 3 compared to Test 1 due to the higher load (50 items per order).
• Unopen Order: Similarly, response times grew approximately 4-5x in Test 3.

Image RemovedImage Added

• The following table compares the average response times for API calls:
  • GET audit-data/acquisition/order/order_id
  • GET audit-data/acquisition/order-line/order_line_id
  • GET audit-data/acquisition/piece/piece_id/status-change-history

Service CPU Utilization

Here we can see that mod-search initiated reindexing for Test #2all modules show a stable trend..

Service Memory Utilization

Here we can see that all modules show a stable trend.

Kafka metrics

OpenSearch Data Nodes metrics

...

DB CPU Utilization

DB CPU in the average were 20% for Scenario 1, 30% for Scenario 2 and 40% for Scenario 3was 12% for all Scenarios.

DB Connections

Max DB connections was 15601250.

DB load

...

PTF - environment Ramsons (rcon)

• 10 m6g.2xlarge EC2 instances located in US East (N. Virginia)us-east-1 [Number of ECS instances, instance type, location region]

• 1 instance of db.r6.xlarge database instance: Writer instance

• OpenSearch

  • domain: ptf-test

  • Number of nodes: 7

  • Version: OpenSearch_2_13_R20240520-P5

• MSK - fse-tenant

  • 4 kafka.m7g.xlarge brokers in 2 zones

  • Apache Kafka version 3.7.x

  • EBS storage volume per broker 300 GiB

  • auto.create.topics.enable=true

  • log.retention.minutes=480

  • default.replication.factor=3

...

• Open Order tub

• Open Order (with 1000 changes in history)

• View Order history

• Open Order Line (with 1000 changes in history)

• View Order Line history

• Open Pieces (with 1000 changes in history)

• View Pieces history

Attached files

Ramsons ECS (rcon) module configuration parameters:

JMeter script:JMeter script: