/
PTF - Performance testing of moving Parsed Records Flattening From Database To Module Space (Quesnelia) [non-ECS]

PTF - Performance testing of moving Parsed Records Flattening From Database To Module Space (Quesnelia) [non-ECS]

Jul 03, 2024

Overview

  • This document contains comparing results of testing  Data Import for MARC Bibliographic records with create and update jobs on the Quesnelia [non-ECS] release on qcp1 environment with two different versions of  mod-source-record-storage 5.8.0and5.9.0-SNAPSHOT.387 (moved json parsing function from DB to module). 

PERF-924: Performance testing of moving Parsed Records Flattening From Database To Module SpaceClosed 

Summary

  • Data Import tests finished successfully on qcp1 environment using the PTF - Create 2 and PTF - Updates Success - 6 profiles with 10k, 100k and 200k file records.

  • DI duration growth correlates to the number of records imported.

  • Comparing between  mod-source-record-storage 5.8.0 and mod-source-record-storage 5.9.0-SNAPSHOT.387

    • Data Import durations have performance degradation of around 35% for all Data-imports jobs.

    • Server and Database metrics at the same level for both versions of mod-source-record-storage module.

  • No memory leaks are observed.

Test Runs and Results

This table contains durations for Data Import. 

Profile

Tenant

MARC File

Test #

DI Duration
mod-source-record-storage
5.8.0
(hh:mm:ss)

Test #

DI Duration
mod-source-record-storage
5.9.0-SNAPSHOT.387
(hh:mm:ss)

Results

Profile

Tenant

MARC File

Test #

DI Duration
mod-source-record-storage
5.8.0
(hh:mm:ss)

Test #

DI Duration
mod-source-record-storage
5.9.0-SNAPSHOT.387
(hh:mm:ss)

Results

DI MARC Bib Create (PTF - Create 2)

fs09000000

25K.mrc

1

0:11:14

7

0:14:30

Completed

fs09000000

100K.mrc

2

0:46:02

8

1:03:37

Completed

fs09000000

200K.mrc

3

1:30:05

9

2:10:52

Completed

DI MARC Bib Update (PTF - Updates Success - 6)

fs09000000

25K.mrc

4

0:37:11

10

0:26:37

Completed

fs09000000

100K.mrc

5

1:21:57

11

1:51:49

Completed

fs09000000

200K.mrc

6

2:43:16

12

3:38:51

Completed

 

Comparison

This table contains durations comparison between  mod-source-record-storage 5.8.0 and mod-source-record-storage 5.9.0-SNAPSHOT.387.

Profile

MARC File

DI Duration
mod-source-record-storage
5.8.0
(hh:mm:ss)

DI Duration
mod-source-record-storage
5.9.0-SNAPSHOT.387
(hh:mm:ss)

Delta
(hh:mm:ss / %)

Profile

MARC File

DI Duration
mod-source-record-storage
5.8.0
(hh:mm:ss)

DI Duration
mod-source-record-storage
5.9.0-SNAPSHOT.387
(hh:mm:ss)

Delta
(hh:mm:ss / %)

DI MARC Bib Create (PTF - Create 2)

25K.mrc

0:11:14

0:14:30

0:03:1
+29%

100K.mrc

0:46:02

1:03:37

0:17:35
+38%

200K.mrc

1:30:05

2:10:52

0:40:47
+45%

DI MARC Bib Update (PTF - Updates Success - 6)

25K.mrc

0:37:11

0:26:37

0:10:34
−28%

100K.mrc

1:21:57

1:51:49

0:29:52
+36%

200K.mrc

2:43:16

3:38:51

0:55:35
+34%

 

 

 

Resource utilization for Test Set №1

CPU

RAM

CPU

RAM

mod-inventory-b

124%

mod-inventory-b

91%

mod-quick-marc-b

92%

mod-source-record-manager-b

71%

mod-di-converter-storage-b

78%

mod-circulation-item-b

58%

nginx-okapi

50%

mod-source-record-storage-b

56%

mod-data-import-b

34%

mod-di-converter-storage-b

53%

mod-source-record-storage-b

27%

okapi-b

44%

okapi-b

24%

mod-data-import-b

44%

mod-source-record-manager-b

20%

mod-patron-blocks-b

44%

mod-inventory-storage-b

18%

mod-feesfines-b

41%

mod-pubsub-b

7%

mod-configuration-b

41%

mod-remote-storage-b

3%

mod-pubsub-b

34%

mod-configuration-b

2%

mod-quick-marc-b

33%

mod-circulation-item-b

2%

mod-users-bl-b

32%

mod-authtoken-b

2%

mod-authtoken-b

29%

mod-feesfines-b

2%

mod-circulation-storage-b

24%

mod-circulation-storage-b

2%

mod-inventory-storage-b

23%

mod-users-bl-b

0.60%

mod-remote-storage-b

21%

mod-patron-blocks-b

0.30%

nginx-okapi

5%

Service CPU Utilization

Here we can see that mod-data-import used 250% CPU in spike and mod-inventory used 115% CPU.

Service Memory Utilization

Here we can see that all modules show a stable trend except mod-source-record-manager.

DB CPU Utilization

DB CPU was 90%.

CPU (User) usage by broker

Disk usage by broker

 

DB Connections

Max number of DB connections was 918.

DB load

Top SQL-queries

 

#

TOP 5 SQL statements

#

TOP 5 SQL statements

1

insert into "marc_records_lb" ("id", "content") values (cast($1 as uuid), cast($2 as jsonb)) on conflict ("id") do update set "content" = cast($3 as jsonb)

2

INSERT INTO fs09000000_mod_source_record_manager.events_processed (handler_id, event_id) VALUES ($1, $2)

3

INSERT INTO fs09000000_mod_source_record_manager.journal_records (id, job_execution_id, source_id, source_record_order, entity_type, entity_id, entity_hrid, action_type, action_status, error,
action_date, title, instance_id, holdings_id, order_id, permanent_location_id, tenant_id) VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11, $12, $13, $14, $15, $16, $17)

4

INSERT INTO fs09000000_mod_inventory_storage.instance (id, jsonb) VALUES ($1, $2) RETURNING jsonb

5

INSERT INTO fs09000000_mod_inventory_storage.item (id, jsonb) VALUES ($1, $2) RETURNING jsonb

Resource utilization for Test Set №2

CPU

RAM

CPU

RAM

mod-inventory-b

111%

mod-inventory-b

92%

mod-quick-marc-b

88%

mod-source-record-manager-b

72%

mod-di-converter-storage-b

60%

mod-circulation-item-b

60%

nginx-okapi

41%

okapi-b

56%

mod-source-record-storage-b

21%

mod-di-converter-storage-b

54%

okapi-b

19%

mod-source-record-storage-b

48%

mod-inventory-storage-b

15%

mod-data-import-b

46%

mod-source-record-manager-b

14%

mod-patron-blocks-b

44%

mod-pubsub-b

7%

mod-feesfines-b

42%

mod-remote-storage-b

3.20%

mod-configuration-b

42%

mod-feesfines-b

2.30%

mod-pubsub-b

36%

mod-configuration-b

2.20%

mod-users-bl-b

35%

mod-circulation-item-b

2.20%

mod-quick-marc-b

34%

mod-circulation-storage-b

1.60%

mod-authtoken-b

31%

mod-data-import-b

1.50%

mod-inventory-storage-b

26%

mod-authtoken-b

1.30%

mod-circulation-storage-b

24%

mod-users-bl-b

0.50%

mod-remote-storage-b

21%

mod-patron-blocks-b

0.30%

nginx-okapi

5%

Service CPU Utilization

Here we can see that mod-data-import used 240% CPU in spike and mod-inventory used 111% CPU.