Resolve API

Performance testing of Resolve API

Approach

10 Million ID-FA Records were created, starting from id value as 1 to id value as 10,000,000. The script to create these records can be found here.

Test script

The test script allows you to adjust the number of records that are submitted in each Resolve Request. All the IDs that are sent in the Resolve request are randomly generated integers (values between 1 and 10,000,000).

This script is then submitted using Locust API requests, where we can specify the number of concurrent users (concurrent API invocations).

The test script can be found here. (TBD)

Readings before test

The following readings were recorded prior to starting the test.

POD	CPU Utilization	Memory Utilization
Mapper POD	0.0225 cpu	477 MB
Postgres POD	0.0080 cpu	20 MB

Test scenarios

8 concurrent users, 1000 requests per API, 1 second ramp up time per user

The following readings were taken 30 minutes after starting the test. The readings were taken while the test was still running.

POD	CPU Utilization	Memory Utilization
Mapper POD	2 cpu	546 MB
Postgres POD	0.25 cpu	1.7 GB

The Locust dashboard at the 30 minute showed the following readings

Locust Dashboard - 30 minutes - 8 concurrent users - 1000 records per API

Locust Charts - 30 minutes - 8 concurrent users - 1000 records per API

The CPU for the Mapper POD showed a 100% throttle at this stage

CPU throttle - 30 minutes - 8 concurrent users - 1000 records per API

4 concurrent users, 1000 requests per API, 1 second ramp up time per user

The following readings were taken 30 minutes after starting the test. The readings were taken while the test was still running.

POD	CPU Utilization	Memory Utilization
Mapper POD	2 cpu	546 MB
Postgres POD	0.225 cpu	1.7 GB

The Locust dashboard at the 30 minute showed the following readings

Locust Dashboard - 30 minutes - 4 concurrent users - 1000 records per API

Locust Charts - 30 minutes - 4 concurrent users - 1000 records per API

The CPU for the Mapper POD showed a 100% throttle at this stage

CPU throttle - 30 minutes - 4 concurrent users - 1000 records per API

4 concurrent users, 2000 requests per API, 1 second ramp up time per user

The following readings were taken 30 minutes after starting the test. The readings were taken while the test was still running.

POD	CPU Utilization	Memory Utilization
Mapper POD	2 cpu	590 MB
Postgres POD	0.235 cpu	1.7 GB

The Locust dashboard at the 30 minute showed the following readings

Locust Dashboard - 30 minutes - 4 concurrent users - 2000 records per API

Locust Charts - 30 minutes - 4 concurrent users - 2000 records per API

The CPU for the Mapper POD showed a 100% throttle at this stage

CPU throttle - 30 minutes - 4 concurrent users - 2000 records per API

The table below summarizes these measurements

Resolve API - Measurements