Infrastructure

PTF -environment ncp3

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

• 2 instances of db.r6.xlarge database instances, one reader, and one writer

• MSK ptf-kakfa-3

  • 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

Modules memory and CPU parameters

MG- Morning Glory release

Front End:

• Item Check-in (folio_checkin-7.2.0)

• Item Check-out (folio_checkout-8.2.0)

Summary 

• In Nolana, there is a significant improvement in the performance of data import and CheckIn/CheckOut compared to KIWI and Lotus.

• The time of Data Import with CI/CO is even better than without CI/CO. Apparently, the reason is for the baseline test the mod-source-record-manager version was 3.5.0 for the test with CI/CO it was 3.5.4.

• From time to time there are spikes in response time for CI/CO up to 10 seconds.

• RDS CPU Utilization was high up to 97%

Results

Bolded numbers are the best response time of CI/CO and the smallest spikes of response time. For  10k MARC Update response time of CI/CO with data import is even better than for CI/CO with 5k MARC Update. For 100k.mrc response time is almost the same as for 10k.mrc that's why it is 2 sets of bolded numbers here.

Comparisons

The following table compares additional test results to previous release numbers and to the CICO baselines Nolana (of Check In average time 0.456s and Checkout average time 0.698s). Note that Lotus numbers are in red, Nolana numbers are in black, and Kiwi numbers are in blue.

In the Nolana version, there is a significant improvement in the performance of data import and CheckIn/CheckOut.

For the baseline test the mod-source-record-manager version was 3.5.0 for the test with CI/CO it was 3.5.4. Apparently, it is the reason why the time of Data Import with CI/CO is even better than without CI/CO.