Overview

• This document contains the results of testing Data Export (MARC BIB) on the Eureka release of Quesnelia FOLIO.  The goal is to compare the performances of Data Export on Eureka and non-Eureka environments and to highlight any observable differences in the KPIs.

PERF-866 - Getting issue details... STATUS  

Summary

• Data Export tests finished successfully on the Eureka environment using the "Default instances export job profile" and "srs - holdings and items job profile." Data Export tests were run on Central tenants.
• When comparing QECP1 Eureka with Quesnelia ECS, NON-ECS environments:
  • • Data Export duration for the "Default instances export job profile" was almost the same across environments.
    • Data Export duration for the "srs - holdings and items job profile" showed a significant degradation: around 2.5 times longer for the 1k and 100k files, and around 5 times longer for the 500k file, and no clear reasons were identified.
• During Tests №1 and №2, we noticed some background processes for Database, so we deactivated mod-fqm-manager. As a result, the repeated test showed a twofold improvement in performance.
• Data Export jobs were getting stuck and returned a [401 Unauthorized] error with the message: "errors [{"type":"UnauthorizedException","code":"authorization_error","message":"Unauthorized"}]. To resolve this, we increased the token lifespan by modifying the KC_CONFIG_TTL parameter for mod-login-keycloak from 360s to 3600s and KC_ADMIN_TOKEN_TTL from 410s to 4100s:
  • • {"name": "KC_CONFIG_TTL","value": "3600s"}
    • {"name": "KC_ADMIN_TOKEN_TTL", "value": "4100s"}

Test Results

This table contains durations for Data Export with 2 job profiles. 

Comparison

This table contains durations comparison between Quesnelia ECS, NON-ECS and Eureka environments.

Service CPU Utilization

Here we can see that mod-data-export used 28% CPU Instance Power in spike.

Service Memory Utilization

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

DB CPU Utilization

DB CPU spike was 68%.

DB Connections

DB connections was 830.

Kafka metrics

OpenSearch Data Nodes metrics

DB load

Top SQL-queries

Instance CPU Utilization

Service CPU Utilization

Here we can see that mod-data-export used 28% CPU Instance Power in spike.

Service Memory Utilization

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

DB CPU Utilization

DB CPU spike was 22%.

DB Connections

DB connections was 850.

Kafka metrics

OpenSearch Data Nodes metrics

DB load

Top SQL-queries

Appendix

Infrastructure

PTF - environment Quesnelia [Eureka] (qecp1)

• 11 m6i.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

Methodology/Approach

Data Export tests scenario using the profiles Default instances export job profile and srs - holdings and items were started from UI on Quesnelia (qecp1) Eureka environment.

Test set:

• Test 1: Manually tested 1k, 100k and 500k records files Data Export started on Main tenant(fs09000000) using Default instances export job profile.
• Test 2: Manually tested 1k, 100k and 500k records files Data Export started on Main tenant(fs09000000) using srs - holdings and items job profile.
• Test 3: Set Instance Count=0 for mod-fqm-manager. Manually tested 1k, 100k and 500k records files Data Export started on Main tenant(fs09000000) using Default instances export job profile.
• Test 4: Set Instance Count=0 for mod-fqm-manager. Manually tested 1k, 100k and 500k records files Data Export started on Main  tenant(fs09000000) using srs - holdings and items job profile.

To get status and time range for export jobs the query used: 

select jsonb->'exportedFiles'->0->>'fileName' as fileName, job_profile_name,exported, started_date,completed_date, completed_date - started_date as duration ,status
from fs09000000_mod_data_export.job_executions where started_date > '2024-10-22' order by started_date desc limit 10;