• Overview

• Summary

• Recommendations & Jira

• Test Runs

• Test Results and Comparison

• Cluster resource utilization for Test 1

  • Service CPU Utilization

  • Service Memory Utilization

  • DB CPU Utilization

  • DB Connections

  • DB Load

  • Top SQL-queries

• Cluster resource utilization for Test 2

  • CICO Response time graph

  • Service CPU Utilization

  • Service Memory Utilization

  • DB CPU Utilization

  • DB Connections

  • DB Load

  • Top SQL-queries

• Resource utilization for Test#3

  • Service CPU Utilization

  • Service Memory Utilization

  • DB CPU Utilization

  • DB Connections

  • DB Load

  • Top SQL-queries

• Appendix

  • Infrastructure

  • Methodology/Approach

This document presents performance testing results for Data Import of MARC Authority records using a create job profile in the Sunflower Eureka non-ECS environments (secp1).

The performance evaluation was carried out across a range of records for a single tenant: 1K, 5K, 10K, 25K, and 50K records. Additionally, we conducted a data import and parallel Check-In/Check-Out test simulating 5 virtual users to assess system behavior under concurrent operations and parallel data import on 3 tenants simultaneously.
Ticket: https://folio-org.atlassian.net/browse/PERF-1110

The report also includes repeated test results executed on the Sunflower Critical Service Patch 1 (CSP1) environment for comparison purposes.
Ticket: https://folio-org.atlassian.net/browse/PERF-1196

• Data import tests finished successfully during Test 1 - Test 2, however in Test 3 only in fs09000000 tenant test finished successfully, but for member tenants failed.

• The data import process during Test 1 of MARC Authority records using a create job profile in the Sunflower release demonstrates a slight but noteworthy improvement in performance compared to the Ramsons release. Our analysis indicates an average speed increase of DI Marc Authority of approximately 16%.

  • The test results indicate that five virtual users (5 VU) for Check-In/Check-Out (CICO) operations do not affect the performance of the data import process.
    
    

• Small files (1K–10K): CSP1 showed slightly slower performance (8–12% longer DI durations).

• Medium and Large files (50K): CSP1 was significantly slower, with DI duration increasing by 8-17%.

• DI Durations: CSP1 generally longer than Sunflower for small to large files

• Check-In/Check-Out (avg times):

  • CI avg increased slightly from 1.08s → 1.23s.

  • CO avg increased from 1.20s → 1.43s.

• Baseline CICO (no DI, Sunflower CSP1):

  • CI: 532 ms (avg), 611 ms (p95).

  • CO: 912 ms (avg), 1297 ms (p95).

• Sunflower: ~26 min (50K) and ~13 min (25K) across all three tenants.

• CSP1: Strong regression — ~43 min (50K) and ~42 min (25K), consistent across tenants.

CSP1 exhibits major performance degradation in multi-tenant imports, with durations nearly doubling compared to baseline.

• During the Data Import Test1, we observed that some job runs failed due to 1–4 records (out of 1000) throwing a “UUID conflict” error.
  To resolve this, we upgraded mod-inventory and its dependent modules to the latest release. This approach successfully unblocked the issue for the central tenant.

  However, the same issue still persists for the member tenant, so further investigation is needed.

• In the Ramsons release, there were issues with mod-permissions on the member tenant fs07-2 during Test3, while testing Data Import concurrently across three tenants.
  We observed that, unlike the previous report release, this time with Sunflower we did not encounter such a blocker. https://folio-org.atlassian.net/browse/PERF-1074

CSP1 . The same set of tests was repeated to validate the system performance after applying Critical Service Patch 1.

Table 1. - Test with 1k, 10k, 25k, and 50k records files DI started on one tenant fs09000000-(secp1-00), and comparative results between Sunflower and Ramsons.

Test №2

Test with CICO 5 concurrent users and DI 1K, 5K, 10K, 25K, and 50K started on one tenant fs09000000-(secp1-00).

Table 2. - Сomparative Baseline Check-In\Check-Out results without Data Import between Ramsons and Sunflower.

Table 3. - Сomparative  Check-In\Check-Out results between Baseline (Sunflower) and  Check-In\Check-Out plus Data Import (Sunflower.)

Test №3
Table 4. - Duration on parallel multitenant data-import on tenants fs09000000-(secp1-00), fs07000001-(secp1-01) and fs07000002-(secp1-02)

Table 2. - Сomparative  Check-In\Check-Out results between Baseline (Sunflower) and  Check-In\Check-Out plus Data Import (Sunflower.)