Data Import with Check-ins Check-outs (Poppy)

Overview

This document contains the results of testing Check-in/Check-out and Data Import for MARC Bibliographic records in the Poppy release.

Ticket: PERF-697 - Getting issue details... STATUS

Summary

There is significant performance improvement for data import in Poppy comparing with Orchid. Durations are closer to Nolana release results. CI/CO response times improved 10% in average with all DI jobs. There's only CI response time degraded in 100k job with 17%.

Comparing Orchid and Poppy releases DI durations in create jobs are up to 10% higher with 5k file and 5% for bigger files with parallel Check-in/Check-out than pure Data import results. DI durations are higher up to 30% in update jobs.

Response times of CI/CO in Poppy release are twice higher with Data Import job with 100k compared with pure CI/CO.

DI create jobs with CI/CO for files 10k, 25k, 50k, 100k perform better in Poppy release and DI update jobs perform better with 25k, 50k, 100k with CI/CO.

No memory leaks are observed.

Average CPU utilization increased for mod-inventory up to 20% comparing with Orchid. So it did not exceed 150% for all the modules. The highest consumption observed from mod-inventory. The rest of services were almost on the same level in the same test in Orchid  Data Import with Check-ins Check-outs Orchid and didn't exceed 60%.

DB needs more connections (in average +20 more) needed for the same tests as in Orchid for all create and update jobs.

Average DB CPU usage is the same as in Orchid - 95%.

Upd: During previous tests on pcp1 and ocp3 there were problems with DI jobs running big files (Create jobs 100k and higher, Update jobs with 25k and higher). The problem was solved after new deployment and updates of 13 modules in scope of ticket  RANCHER-1121 - Getting issue details... STATUS and RANCHER-1114 - Getting issue details... STATUS and the large DI jobs are completing successfully now.


Test Runs 

Test #

Scenario

Load levelComment
1DI MARC Bib Create5K, 10K, 25K, 50K, 100K consequentially (with 5 min pause)
CICO 8 users
2DI MARC Bib Update5K, 10K, 25K, 50K, 100K consequentially (with 5 min pause)
CICO 8 users

Test Results

Data import

Total time for all Data Export jobs - 1 hour 16 minutes 47 seconds.

Profile
MARC File


DI Duration

Poppy (hh:mm:ss)

Check In, Check Out Response time (8 users)

Poppy

CI Average secCO Average sec
DI MARC Bib Create (PTF - Create 2)5K.mrc00:02:530.9011.375
10K.mrc00:04:320.9021.47
25K.mrc00:11:1411.571
50K.mrc00:21:550.9811.46
100K.mrc00:47:021.0181.491
Data Export MARC Bib (Export for Data Import updates)



5K.mrc00:02:090.4950.836
10K.mrc00:04:190.4680.917
25K.mrc00:10:300.4970.935
50K.mrc00:20:110.5090.923
100K.mrc00:39:38

DI MARC Bib Update (PTF - Updates Success - 1)5K.mrc00:03:190.7551.169
10K.mrc00:06:200.751.307
25K.mrc00:14:040.8221.403
50K.mrc00:29:590.8931.424
100K.mrc01:03:030.9081.51

Check-in/Check-out without DI

ScenarioLoad levelRequestResponse time, sec
Poppy
95 percaverage
Circulation Check-in/Check-out (without Data import)8 usersCheck-in0.4890.431
Check-out0.9690.828

Comparison

CICO with DI comparison

ProfileMARC FileDI DurationDeviation, % (compared DI Poppy without CICO and with CICO)DI Delta, (hh:mm:ss) Orchid/PoppyCheck In, Check Out Response time (8 users)Check In, Check Out Response time (8 users)Delta, %

without CI/COwith CI/COOrchidPoppyOrchid/PoppyOrchid/Poppy
Orchid*PoppyOrchid*PoppyCI Average secCO Average secCI Average secCO Average secCICO
DI MARC Bib Create (PTF - Create 2)5K.mrc00:04:3000:02:3900:05:0100:02:53+8.5% / 14 sec - 00:02:080.9611.4420.9011.375-6.24%-4.65%
10K.mrc00:09:2500:05:0000:09:0600:04:32-9.3% / 28 sec- 00:04:351.0581.6240.9021.47-14.74%-9.48%
25K.mrc00:22:1600:11:1500:24:2800:11:14-0.2% / 1 sec- 00:13:141.0561.62111.571-5.30%-3.08%
50K.mrc00:39:2700:22:1600:43:0300:21:55-1.5% / 21 sec- 00:21:090.9361.5190.9811.464.81%-3.8