Skip to end of banner
Go to start of banner

MaxRAMPercentage vs Xmx with Data Import and Check-ins Check-outs (Poppy)

Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 3 Next »

Overview IN PROGRES 

This document contains the results of testing different configurations of modules  with changes in JAVA_OPTS replacing XMX with MaxRAMPercentage. The main KPI for these tests are Memory consumption of modules, Data import durations and Check-in / Check-out response times.

Ticket: PERF-893 - Getting issue details... STATUS

Summary

  • Replacing XMX with MaxRAMPercentage do not affect main KPI's. Maximum delta of DI's durations  was close to 4 minutes that is acceptable.
  • Changing MaxRAMPercentage from default values to recommended 66% do not affect Check-in / Check-out response times but it affect Data import durations more than 5 minutes difference. For DI test #6 - almost 8 minutes.
  • Data import perform better in configuration #2. Comparing DI durations between configuration #2 and #3 (with limited MaxRAMPercentage) we see 4 minutes difference. Problems with mod-inventory module appear in configuration #3 faster.
  • Memory consumption from modules 

Test Runs 

Test / Configuration #

Scenario

Load levelJAVA_OPTS
120 vUsers longevity CI/CO + DI MARC Bib Create12 Data Import tests with 100K sequentially (with 5 min pause)
-XMX
220 vUsers longevity CI/CO + DI MARC Bib Create12 Data Import tests with 100K sequentially (with 5 min pause)
-XX:MaxRAMPercentage, Default
320 vUsers longevity CI/CO + DI MARC Bib Create12 Data Import tests with 100K sequentially (with 5 min pause)-XX:MaxRAMPercentage, Recommended - 66%

Test Results

Comparison

The following table compares test results 

Configuration #1 BASELINE, XMX   Configuration #2 default MaxRAMPercentage

Configuration #3 recommended MaxRAMPercentage, 66%
16 hours CICOTest #hh:mm:ss24 hours CI/COTest #hh:mm:ssdelta, default - xmx 24 hours CI/COTest #hh:mm:ssdelta, 66% - xmx








DI 100k Create #1200:56:32These results after 1 container for mod-inventory stopped.
A new 1 container was created.




DI 100k Create #1101:39:094 1k jobs stopped
DI 100k Create #1100:55:58




DI 100k Create #1001:03:19

DI 100k Create #1000:57:00




DI 100k Create #900:59:40

DI 100k Create #900:56:10

DI 100k Create #800:55:01
DI 100k Create #800:58:4700:03:46
DI 100k Create #800:55:23

DI 100k Create #700:54:43
DI 100k Create #700:58:0800:03:26
DI 100k Create #702:46:273 1k jobs stopped

DI 100k Create #600:55:25
DI 100k Create #600:56:0900:00:44
DI 100k Create #601:03:0700:07:42

DI 100k Create #500:54:45
DI 100k Create #500:58:1900:03:33
DI 100k Create #501:03:3700:08:52

DI 100k Create #400:55:38
DI 100k Create #400:54:2700:01:11
DI 100k Create #400:59:1100:03:33

DI 100k Create #300:53:30
DI 100k Create #300:54:3600:01:06
DI 100k Create #300:58:3400:05:04

DI 100k Create #200:53:16
DI 100k Create #200:53:0300:00:13
DI 100k Create #200:57:5900:04:43

DI 100k Create #100:52:51
DI 100k Create #101:05:1000:12:18
DI 100k Create #101:05:4200:12:50











 Results taken from test #8DI +DI -
DI +DI -

DI +DI -
CI1220538CI1078486
CI1240470
CO2311978CO19671041
CO2121965

Detailed CICO response time for CICO with DI

Request*Response time (avg, sec)
Pure CICOCICO + 100K MARC BIB Create




































*Top-10 requests were taken for analysis.

Response time

DI MARC BIB Create + CICO



Service CPU Utilization


DI MARC BIB Update+ CICO


Service Memory Utilization


DI MARC BIB Create + CICO



DB CPU Utilization

DI MARC BIB Create + CICO


DB Connections

DI MARC BIB Create + CICO



DB load

DI MARC BIB Create + CICO



Top SQL-queries:


Appendix

Infrastructure

PTF -environment pcp1

  • 10 m6i.2xlarge EC2 instances located in US East (N. Virginia)us-east-1

  • 2 database  instances, writer/reader

    NameMemory GIBvCPUsmax_connections

    db.r6g.xlarge

    		32 GiB4 vCPUs2731
  • MSK tenant
    • 4 m5.2xlarge brokers in 2 zones

    • Apache Kafka version 2.8.0

    • EBS storage volume per broker 300 GiB

    • auto.create.topics.enable=true
    • log.retention.minutes=480
    • default.replication.factor=3
ModuleTask Def. RevisionTask Def. Revision with MaxRAMPercentage = by defaultTask Def. Revision with MaxRAMPercentage = 66.0Module VersionTask CountMem Hard LimitMem Soft limitCPU unitsXmxMetaspaceSizeMaxMetaspaceSizeR/W split enabledMaxRAMPercentage
Tests 		MaxRAMPercentage = recommended (66%)
pcp1-pvt













Thu Apr 04 09:40:28 UTC 2024













mod-data-import202324mod-data-import:3.0.71204818442561292384512FALSE70.166.0
mod-authtoken161718mod-authtoken:2.14.121440115251292288128FALSE80.066.0
mod-configuration101213mod-configuration:5.9.22102489612876888128FALSE85.766.0
mod-users-bl101112mod-users-bl:7.6.021440115251292288128FALSE80.066.0
mod-inventory-storage151718mod-inventory-storage:27.0.424096369020483076384512FALSE83.466.0
mod-circulation-storage141516mod-circulation-storage:17.1.722880259215361814384512FALSE70.066.0
mod-source-record-storage181920mod-source-record-storage:5.7.525600500020483500384512FALSE70.066.0
mod-inventory151617mod-inventory:20.1.822880259210241814384512FALSE70.066.0
mod-di-converter-storage181920mod-di-converter-storage:2.1.52102489612876888128FALSE85.766.0
mod-circulation141516mod-circulation:24.0.1122880259215361814384512FALSE70.066.0
mod-pubsub131415mod-pubsub:2.11.32153614401024922384512FALSE64.066.0
mod-users343536mod-users:19.3.0-SNAPSHOT.6772102489612876888128FALSE85.766.0
mod-patron-blocks101112mod-patron-blocks:1.9.021024896102476888128FALSE85.766.0
mod-source-record-manager171920mod-source-record-manager:3.7.825600500020483500384512FALSE70.066.0
nginx-okapi9

nginx-okapi:2023.06.1421024896128000FALSE0.0
okapi-b111213okapi:5.1.23168414401024922384512FALSE64.066.0
mod-feesfines111213mod-feesfines:19.0.02102489612876888128FALSE85.766.0

Methodology/Approach

  1. Run baseline for Check-in / Check-out longevity test with Data import Create job with 100k file.
  2. Change revisions of modules with needed configuration of JAVA_OPTS - replaced XMX with MaxRAMPercentage in aa
  3. Run CI/CO
  4. Run DI script after ramp-up period for CI/CO or at least 10-15 minutes

Additional information


  • No labels