Back
— Feb 27, 2025 ·

Comparing VPS Performance Across Budget Options

This journey began a little while ago. I've been running a VPS at Contabo for years, using Ubuntu with Virtualmin to manage my projects. But I was not really happy with the setup. More and more of my projects became NextJS or Laravel based. For NextJS, I would typically use Vercel and their hobby plan, while Laravel was manageable on my VPS. For automatic deployments, I had a GitHub Action SSH-ing into the machine to deploy. Here's an example deployment script I was using:

          name: Laravel Deploy

on:
  push:
    branches:
      - main

jobs:
  deploy:
    runs-on: ubuntu-latest

    steps:
      - name: Checkout Repository
        uses: actions/checkout@v2

      - name: Configure SSH
        run: |
          mkdir -p ~/.ssh/
          echo "${{ secrets.APIKEY }}" > ~/.ssh/id_rsa
          chmod 600 ~/.ssh/id_rsa
          ssh-keyscan -H "app.example.com" >> ~/.ssh/known_hosts

      - name: Set up PHP
        uses: shivammathur/setup-php@v2
        with:
          php-version: "8.2"
          extensions: mbstring, intl, bcmath

      - name: Install Composer Dependencies
        run: composer install --no-interaction --prefer-dist --optimize-autoloader --no-dev
        env:
          AWS_ACCESS_KEY_ID: "${{ secrets.AWS_ACCESS_KEY_ID }}"
          AWS_SECRET_ACCESS_KEY: "${{ secrets.AWS_SECRET_ACCESS_KEY }}"
          AWS_DEFAULT_REGION: "eu2"
          AWS_BUCKET: "bucket"
          AWS_ENDPOINT: "https://eu2.contabostorage.com/"
          AWS_URL: "https://contabobucketurl.com/"
          AWS_USE_PATH_STYLE_ENDPOINT: "true"

      - name: Set up Node.js
        uses: actions/setup-node@v2
        with:
          node-version: "21"

      - name: Install NPM Dependencies
        run: npm install
        env:
          PUPPETEER_SKIP_DOWNLOAD: true

      - name: Build Assets
        run: npm run build

      - name: Deploy
        run: |
          rsync -avz --delete --exclude=storage/app/public/folder/ --exclude=.env --exclude=auth.json --itemize-changes --stats -e "ssh -i ~/.ssh/id_rsa" ./ [email protected]:/home/app/public_html
          ssh -i ~/.ssh/id_rsa [email protected] 'cd /home/app/public_html && php artisan migrate --force && php artisan config:cache && php artisan storage:link && npx puppeteer browsers install chrome && cp -r .cache/ ../ && php artisan queue:restart'

      - name: Clear SSH key
        run: rm ~/.ssh/id_rsa

This worked, but it wasn't ideal. There were no rolling deployments, so if something went wrong, my app would go down. Sure, these are all hobby projects, but some people do use them, and I don't like troubleshooting an app that's down.

The Solution: Coolify

I decided to move to Coolify, an open-source alternative to Vercel and Netlify which can be self-hosted. Coolify provides a more robust deployment pipeline with features like rollbacks and better monitoring. I wanted to transition gradually, so I got myself the cheapest VPS at Contabo, but quickly ran into memory limitations. Running 12GB swap on a 4GB RAM VPS is not exactly best practice.

Processor  : AMD EPYC 9555P 64-Core Processor
CPU cores  : 4 @ 3195.200 MHz
RAM        : 3.8 GiB
Swap       : 12.0 GiB
Disk       : 96.8 GiB
Distro     : Ubuntu 24.04.2 LTS
Kernel     : 6.8.0-53-generic

This led me to look around for alternatives, and I found Layer7, which offers some very affordable options on paper. I quickly set up a VPS there with the following specs:

Processor  : Intel(R) Xeon(R) Gold 6230 CPU @ 2.10GHz
CPU cores  : 6c @ 2.100 Ghz
RAM        : 15.6 GB RAM (0.0 MB SWAP)
Disk.      : 118.0 GB Disk
Distro:    : Ubuntu 24.04.2 LTS
Kernel     : 6.8.0-53-generic

I quickly realized that this machine was significantly slower than my 4-core 4GB RAM Contabo VPS. Build times for the same app took more than twice as long. In one of my current projects, PoopLoop, I noticed a substantial performance difference in forms that use conditional logic with Livewire.

The Benchmark Test

This experience motivated me to run some proper benchmarks. I found this repository for running standardized tests across different servers. I decided to test several VPS options, focusing on the €10-15 price range since these are hobby projects.
Here are the VPS configurations I tested:

Provider Model CPU RAM Price (€/month)
Contabo (old) VPS 6 cores 16GB ~€12
Contabo (new) VPS 4 cores 4GB ~€6
Layer7 Intel VPS 6 cores 16GB ~€12
Layer7 AMD VPS 6 cores 16GB ~€12
Layer7 AMD VDS 4 cores 16GB ~€16

Note: VPS performance is always dependent on your neighbors. These benchmarks represent performance at the time of testing only.

Benchmark results

For your reading, the pages to the results:

  1. My old vps at contabo 6c 16GB ram Geekbench Yabs
  2. My new vps at contabo 4c 4Gb ram Geekbench Yabs
  3. My new intel vps at Layer7 6c 16GB ram Geekbench Yabs
  4. A new AMD vps at Layer7 6c 16GB ram Geekbench Yabs
  5. A new AMD vds at Layer7 4v 16GB ram Geekbench Yabs

CPU Performance Comparison

The Geekbench CPU benchmark was the most important metric for my use case, as it directly impacts build times and application performance. Here's how the systems compared:

Provider Model Single-Core Score Multi-Core Score
Contabo (old) VPS 811 2.560
Contabo (new) VPS 2.279 7.419
Layer7 Intel VPS 716 2.480
Layer7 AMD VPS 1.199 5.067
Layer7 AMD VDS 1,204 3,835

The Contabo VPS options clearly outperformed the Layer7 alternatives in CPU benchmarks, which explains why I was seeing significantly slower build times on the Layer7 VPS.

My Current Workload

To give context to why performance matters, here's what I'm running on my VPS with Coolify:

  1. Two NextJS front-ends with external databases
  2. A NextJS front-end with WordPress back-end in the same project
  3. A relatively popular WordPress blog (HugoKookt.com)
  4. A plain PHP application
  5. Four Laravel applications with MySQL databases
  6. Glances for some container monitoring
  7. In the future probably plausible analytics

With this workload, CPU performance for builds and sufficient RAM for multiple containers are essential for smooth operation.

Conclusion and Decision

After analyzing the benchmark results, there's a clear difference between the new Contabo VPS and the alternatives. The Contabo VPS offers significantly better CPU performance, which is critical for my build processes and application responsiveness.

Based on these findings, I've decided to stick with the new Contabo VPS but upgrade it to have more RAM. The base 4GB RAM model at ~€6/month provides excellent CPU performance, but I need additional memory to comfortably run all my containers. The upgraded model remains the most cost-effective option in my €10-15 budget range. (and here is the Yabs result)

Of course, VPS performance can vary over time as neighbors change, but the benchmarks provide a solid baseline for decision-making.

Copyright © Hugo Winder. All rights reserved.