I have a spring boot rest api with a POST endpoint. This is a write heavy endpoint which does 10-15 entries into 2-4 tables (business requirement). So I created a docker-compose.yml for load/stress testing the application using Gatling. I made 2 instances of Postgres, one is master and one is replica. Now to load test the api I assigned different cpu and memory to the service. Below are the results.

1. For 1 cpu and 1024MB memory: The api works best with 70 constant request per second. All requests are under 800ms (Gatling graph) with minimum 15ms and max 75ms response time. The DB cpu goes to 40% while the api cpu utilisation goes upto 105%. Memory usage is 800MB. If I increase the req/sec I start getting letancy.

2. For 2/3/4 cpu and 1024MB memory: cpu utilisation goes to 200/300/400% respectivly and still the 70req/sec is the best the server can handle. If I increase req/sec the same latency comes. DB and memory utilisation is same.

So my question is, shouldn’t I get better results with more cpu? Why my throughput is always 70req/sec irrespective of the number the CPUs I provide?

Spring boot version: 3.2.5

OS: MacOs, 32Gb, 10 cpu core

docker-compose.yml

version: "3.9"

x-postgres-common:
  &postgres-common
  image: postgres:13
  restart: always
  user: postgres

services:
  postgres_primary:
    <<: *postgres-common
    ports:
      - "5440:5432"
    environment:
      POSTGRES_USER: user
      POSTGRES_DB: cdfsegbdb
      POSTGRES_PASSWORD: password
      POSTGRES_HOST_AUTH_METHOD: "scram-sha-256nhost replication all 0.0.0.0/0 md5"
      POSTGRES_INITDB_ARGS: "--auth-host=scram-sha-256"
    command: |
      postgres 
      -c wal_level=replica 
      -c hot_standby=on 
      -c max_wal_senders=10 
      -c max_replication_slots=10 
      -c hot_standby_feedback=on
    volumes:
      - ./src/test/resources/db/init.sql:/docker-entrypoint-initdb.d/00_init.sql
    networks:
      - cdfs-network

  postgres_replica:
    <<: *postgres-common
    ports:
      - "5441:5432"
    environment:
      POSTGRES_USER: user
      POSTGRES_DB: cdfsegbdb
      POSTGRES_PASSWORD: password
      PGUSER: replicator
      PGPASSWORD: replicator_password
    command: |
      bash -c "
      rm -rf /var/lib/postgresql/data/*
      until pg_basebackup --pgdata=/var/lib/postgresql/data -R --slot=replication_slot --host=cdfs_postgres_primary --port=5432;
      do
      echo 'Waiting for primary to connect...'
      sleep 1s
      done
      echo 'Backup done, starting replica...'
      chmod 0700 /var/lib/postgresql/data
      postgres
      "
    depends_on:
      - postgres_primary

  
  my-api-service:
    image: my-api:latest
    ports:
      - "8080:8080"
    depends_on:
      - postgres_primary
      - postgres_replica
    deploy:
      resources:
        limits:
          cpus: 4
          memory: 2048M
        reservations:
          cpus: 4
          memory: 2048M