The latest articles on DEV Community by Adrian Silaghi (@adrian_silaghi_7b8ec8d2f7). https://dev.to/adrian_silaghi_7b8ec8d2f7 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3658977%2F911f169b-8b5e-439a-a68d-7c3e1b1ec55e.jpeg https://dev.to/adrian_silaghi_7b8ec8d2f7 en Adrian Silaghi Fri, 12 Dec 2025 19:40:27 +0000 https://dev.to/adrian_silaghi_7b8ec8d2f7/vps-benchmark-danubedata-vs-digitalocean-the-numbers-dont-lie-20ff https://dev.to/adrian_silaghi_7b8ec8d2f7/vps-benchmark-danubedata-vs-digitalocean-the-numbers-dont-lie-20ff <p>Here is the content converted to Markdown.</p> <p>When choosing a cloud provider, benchmarks speak louder than marketing claims. We ran the industry-standard <a href="https://github.com/masonr/yet-another-bench-script" rel="noopener noreferrer">YABS (Yet-Another-Bench-Script)</a> on VPS instances from DanubeData and DigitalOcean to see how they stack up.</p> <h2> The Test Setup </h2> <p>We tested three configurations, all running Ubuntu 24.04 LTS in German datacenters:</p> <ul> <li> <strong>DanubeData DD Nano:</strong> 2 vCPU, 2GB RAM, 40GB NVMe - Falkenstein, Germany - <strong>€8.99/month</strong> </li> <li> <strong>DigitalOcean Premium AMD:</strong> 2 vCPU, 2GB RAM, 60GB SSD - Frankfurt, Germany - <strong>$21/month</strong> </li> <li> <strong>DigitalOcean CPU-Optimized Intel:</strong> 2 vCPU, 4GB RAM, 25GB SSD - Frankfurt, Germany - <strong>$42/month</strong> </li> </ul> <p>DigitalOcean's CPU-Optimized tier costs <strong>nearly 5x more</strong> than DanubeData and promises dedicated CPU threads for compute-intensive workloads. Let's see if it delivers.</p> <h2> CPU Performance: Geekbench 6 Results </h2> <p>The results speak for themselves:</p> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Instance</th> <th>Price</th> <th>Single Core</th> <th>Multi Core</th> </tr> </thead> <tbody> <tr> <td><strong>DanubeData DD Nano</strong></td> <td><strong>€8.99</strong></td> <td><strong>2,762</strong></td> <td><strong>4,913</strong></td> </tr> <tr> <td>DO Premium AMD</td> <td>$21</td> <td>1,282</td> <td>2,303</td> </tr> <tr> <td>DO CPU-Optimized Intel</td> <td>$42</td> <td>1,098</td> <td>1,308</td> </tr> </tbody> </table></div> <p>DanubeData's €8.99 instance delivers:</p> <ul> <li> <strong>+115%</strong> better single-core performance vs DO Premium AMD ($21)</li> <li> <strong>+151%</strong> better single-core performance vs DO CPU-Optimized Intel ($42)</li> <li> <strong>+275%</strong> better multi-core performance vs DO CPU-Optimized Intel ($42)</li> </ul> <p>Yes, you read that correctly. DigitalOcean's "CPU-Optimized" tier running an Intel Xeon Platinum 8168 at 2.7 GHz gets crushed by DanubeData's AMD EPYC Genoa at 3.79 GHz—despite costing nearly 5x more.</p> <h2> Disk I/O: Local NVMe Dominance </h2> <p>Storage performance shows an even more dramatic difference:</p> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Test (Mixed R/W)</th> <th>DanubeData</th> <th>DO Premium AMD</th> <th>DO CPU-Optimized</th> </tr> </thead> <tbody> <tr> <td>4K Total</td> <td><strong>329 MB/s (82K IOPS)</strong></td> <td>202 MB/s (50K IOPS)</td> <td>237 MB/s (59K IOPS)</td> </tr> <tr> <td>64K Total</td> <td><strong>3.26 GB/s (51K IOPS)</strong></td> <td>1.19 GB/s (18K IOPS)</td> <td>330 MB/s (5K IOPS)</td> </tr> <tr> <td>512K Total</td> <td><strong>3.21 GB/s</strong></td> <td>1.08 GB/s</td> <td>452 MB/s</td> </tr> <tr> <td>1M Total</td> <td><strong>2.57 GB/s</strong></td> <td>632 MB/s</td> <td>642 MB/s</td> </tr> </tbody> </table></div> <p>DanubeData's local NVMe storage delivers:</p> <ul> <li> <strong>10x faster</strong> 64K throughput vs DO CPU-Optimized (3.26 GB/s vs 330 MB/s)</li> <li> <strong>7x faster</strong> 512K throughput vs DO CPU-Optimized (3.21 GB/s vs 452 MB/s)</li> <li> <strong>4x faster</strong> 1M throughput vs DO Premium AMD (2.57 GB/s vs 632 MB/s)</li> </ul> <h3> Why Local NVMe Matters </h3> <p>Most cloud providers use network-attached storage (NAS/SAN) to share storage across physical hosts. While this enables features like live migration, it adds latency to every I/O operation. DanubeData uses direct-attached enterprise NVMe drives (Samsung PM9A3, Intel D7-P5520) connected via PCIe 4.0, eliminating network overhead entirely.</p> <h2> Network Performance </h2> <p>DigitalOcean shows higher raw network bandwidth in the iperf3 tests, with speeds up to 16 Gbits/sec receive to nearby locations. DanubeData's network caps at around 930 Mbits/sec. However, consider that:</p> <ul> <li>1 Gbps is sufficient for the vast majority of workloads</li> <li>DanubeData's internal traffic between VPS, databases, and caches is <strong>free and sub-millisecond</strong> within the same namespace</li> <li>For bandwidth-intensive applications, the focus should be on your actual requirements</li> </ul> <h2> The Hardware Comparison </h2> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Specification</th> <th>DanubeData</th> <th>DO Premium AMD</th> <th>DO CPU-Optimized</th> </tr> </thead> <tbody> <tr> <td>Processor</td> <td><strong>AMD EPYC Genoa (Zen4)</strong></td> <td>DO-Premium-AMD</td> <td>Xeon Platinum 8168</td> </tr> <tr> <td>Clock Speed</td> <td><strong>3,792 MHz</strong></td> <td>2,299 MHz</td> <td>2,693 MHz</td> </tr> <tr> <td>Architecture</td> <td><strong>Zen4 (2022)</strong></td> <td>Unknown</td> <td>Skylake (2017)</td> </tr> <tr> <td>Storage Type</td> <td><strong>Local NVMe (PCIe 4.0)</strong></td> <td>Network SSD</td> <td>Network SSD</td> </tr> <tr> <td>Memory</td> <td><strong>DDR5</strong></td> <td>DDR4</td> <td>DDR4</td> </tr> <tr> <td>YABS Runtime</td> <td><strong>8 min 53 sec</strong></td> <td>11 min 36 sec</td> <td>13 min 48 sec</td> </tr> </tbody> </table></div> <p>DigitalOcean's "CPU-Optimized" tier runs on Intel Xeon Platinum 8168 processors from 2017—that's 5-year-old silicon at 2025 prices. DanubeData runs AMD EPYC Genoa, the latest generation of datacenter processors.</p> <h2> Price-Performance: The Complete Picture </h2> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Metric</th> <th>DanubeData</th> <th>DO Premium AMD</th> <th>DO CPU-Optimized</th> </tr> </thead> <tbody> <tr> <td>Monthly Price</td> <td><strong>€8.99</strong></td> <td>$21</td> <td>$42</td> </tr> <tr> <td>Geekbench Single-Core</td> <td><strong>2,762</strong></td> <td>1,282</td> <td>1,098</td> </tr> <tr> <td>Points per Dollar*</td> <td><strong>~280</strong></td> <td>61</td> <td>26</td> </tr> </tbody> </table></div> <p>&lt;small&gt;*DanubeData converted at ~€1 = $1.05 for comparison&lt;/small&gt;</p> <p>DanubeData delivers:</p> <ul> <li> <strong>4.6x</strong> better price-performance than DO Premium AMD</li> <li> <strong>10.8x</strong> better price-performance than DO CPU-Optimized Intel</li> </ul> <h2> What This Means for Your Workloads </h2> <ul> <li> <strong>Databases:</strong> Up to 10x faster disk I/O means faster queries, quicker backups, and better transaction throughput</li> <li> <strong>Compilation &amp; Build:</strong> 2-3x CPU performance cuts build times dramatically</li> <li> <strong>Web Applications:</strong> Faster response times and higher concurrency</li> <li> <strong>Data Processing:</strong> Process datasets faster with modern Zen4 architecture</li> <li> <strong>Cost Efficiency:</strong> Get better performance while saving up to 79% on monthly costs</li> </ul> <h2> Reproducible Results </h2> <p>These benchmarks are fully reproducible. Spin up a DanubeData VPS and run the test yourself:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight shell"><code>curl <span class="nt">-sL</span> yabs.sh | bash </code></pre> </div> <p>Full Geekbench results:</p> <ul> <li><a href="https://browser.geekbench.com/v6/cpu/15429256" rel="noopener noreferrer">DanubeData DD Nano - Geekbench Result</a></li> <li><a href="https://browser.geekbench.com/v6/cpu/15429317" rel="noopener noreferrer">DigitalOcean Premium AMD - Geekbench Result</a></li> <li><a href="https://browser.geekbench.com/v6/cpu/15429932" rel="noopener noreferrer">DigitalOcean CPU-Optimized Intel - Geekbench Result</a></li> </ul> <h2> The Bottom Line </h2> <p>DigitalOcean's "CPU-Optimized" tier at $42/month runs on 2017-era Intel Xeon processors and gets outperformed by DanubeData's €8.99 entry-level VPS by a factor of 2.5x in CPU benchmarks and up to 10x in storage throughput.</p> <p>This isn't about minor optimizations—it's about running modern hardware. AMD EPYC Genoa (Zen4) with DDR5 and local NVMe storage simply operates in a different performance class than legacy infrastructure.</p> <p>Ready to experience the difference? <a href="https://danubedata.ro/register" rel="noopener noreferrer">Deploy your first VPS</a> and run the benchmarks yourself.</p> <h2> * Disclosure: This benchmark was conducted by DanubeData. We encourage you to run your own tests.* </h2> devops cloud vps benchmark Adrian Silaghi Fri, 12 Dec 2025 19:40:27 +0000 https://dev.to/adrian_silaghi_7b8ec8d2f7/vps-benchmark-danubedata-vs-digitalocean-the-numbers-dont-lie-pfn https://dev.to/adrian_silaghi_7b8ec8d2f7/vps-benchmark-danubedata-vs-digitalocean-the-numbers-dont-lie-pfn <p>Here is the content converted to Markdown.</p> <p>When choosing a cloud provider, benchmarks speak louder than marketing claims. We ran the industry-standard <a href="https://github.com/masonr/yet-another-bench-script" rel="noopener noreferrer">YABS (Yet-Another-Bench-Script)</a> on VPS instances from DanubeData and DigitalOcean to see how they stack up.</p> <h2> The Test Setup </h2> <p>We tested three configurations, all running Ubuntu 24.04 LTS in German datacenters:</p> <ul> <li> <strong>DanubeData DD Nano:</strong> 2 vCPU, 2GB RAM, 40GB NVMe - Falkenstein, Germany - <strong>€8.99/month</strong> </li> <li> <strong>DigitalOcean Premium AMD:</strong> 2 vCPU, 2GB RAM, 60GB SSD - Frankfurt, Germany - <strong>$21/month</strong> </li> <li> <strong>DigitalOcean CPU-Optimized Intel:</strong> 2 vCPU, 4GB RAM, 25GB SSD - Frankfurt, Germany - <strong>$42/month</strong> </li> </ul> <p>DigitalOcean's CPU-Optimized tier costs <strong>nearly 5x more</strong> than DanubeData and promises dedicated CPU threads for compute-intensive workloads. Let's see if it delivers.</p> <h2> CPU Performance: Geekbench 6 Results </h2> <p>The results speak for themselves:</p> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Instance</th> <th>Price</th> <th>Single Core</th> <th>Multi Core</th> </tr> </thead> <tbody> <tr> <td><strong>DanubeData DD Nano</strong></td> <td><strong>€8.99</strong></td> <td><strong>2,762</strong></td> <td><strong>4,913</strong></td> </tr> <tr> <td>DO Premium AMD</td> <td>$21</td> <td>1,282</td> <td>2,303</td> </tr> <tr> <td>DO CPU-Optimized Intel</td> <td>$42</td> <td>1,098</td> <td>1,308</td> </tr> </tbody> </table></div> <p>DanubeData's €8.99 instance delivers:</p> <ul> <li> <strong>+115%</strong> better single-core performance vs DO Premium AMD ($21)</li> <li> <strong>+151%</strong> better single-core performance vs DO CPU-Optimized Intel ($42)</li> <li> <strong>+275%</strong> better multi-core performance vs DO CPU-Optimized Intel ($42)</li> </ul> <p>Yes, you read that correctly. DigitalOcean's "CPU-Optimized" tier running an Intel Xeon Platinum 8168 at 2.7 GHz gets crushed by DanubeData's AMD EPYC Genoa at 3.79 GHz—despite costing nearly 5x more.</p> <h2> Disk I/O: Local NVMe Dominance </h2> <p>Storage performance shows an even more dramatic difference:</p> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Test (Mixed R/W)</th> <th>DanubeData</th> <th>DO Premium AMD</th> <th>DO CPU-Optimized</th> </tr> </thead> <tbody> <tr> <td>4K Total</td> <td><strong>329 MB/s (82K IOPS)</strong></td> <td>202 MB/s (50K IOPS)</td> <td>237 MB/s (59K IOPS)</td> </tr> <tr> <td>64K Total</td> <td><strong>3.26 GB/s (51K IOPS)</strong></td> <td>1.19 GB/s (18K IOPS)</td> <td>330 MB/s (5K IOPS)</td> </tr> <tr> <td>512K Total</td> <td><strong>3.21 GB/s</strong></td> <td>1.08 GB/s</td> <td>452 MB/s</td> </tr> <tr> <td>1M Total</td> <td><strong>2.57 GB/s</strong></td> <td>632 MB/s</td> <td>642 MB/s</td> </tr> </tbody> </table></div> <p>DanubeData's local NVMe storage delivers:</p> <ul> <li> <strong>10x faster</strong> 64K throughput vs DO CPU-Optimized (3.26 GB/s vs 330 MB/s)</li> <li> <strong>7x faster</strong> 512K throughput vs DO CPU-Optimized (3.21 GB/s vs 452 MB/s)</li> <li> <strong>4x faster</strong> 1M throughput vs DO Premium AMD (2.57 GB/s vs 632 MB/s)</li> </ul> <h3> Why Local NVMe Matters </h3> <p>Most cloud providers use network-attached storage (NAS/SAN) to share storage across physical hosts. While this enables features like live migration, it adds latency to every I/O operation. DanubeData uses direct-attached enterprise NVMe drives (Samsung PM9A3, Intel D7-P5520) connected via PCIe 4.0, eliminating network overhead entirely.</p> <h2> Network Performance </h2> <p>DigitalOcean shows higher raw network bandwidth in the iperf3 tests, with speeds up to 16 Gbits/sec receive to nearby locations. DanubeData's network caps at around 930 Mbits/sec. However, consider that:</p> <ul> <li>1 Gbps is sufficient for the vast majority of workloads</li> <li>DanubeData's internal traffic between VPS, databases, and caches is <strong>free and sub-millisecond</strong> within the same namespace</li> <li>For bandwidth-intensive applications, the focus should be on your actual requirements</li> </ul> <h2> The Hardware Comparison </h2> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Specification</th> <th>DanubeData</th> <th>DO Premium AMD</th> <th>DO CPU-Optimized</th> </tr> </thead> <tbody> <tr> <td>Processor</td> <td><strong>AMD EPYC Genoa (Zen4)</strong></td> <td>DO-Premium-AMD</td> <td>Xeon Platinum 8168</td> </tr> <tr> <td>Clock Speed</td> <td><strong>3,792 MHz</strong></td> <td>2,299 MHz</td> <td>2,693 MHz</td> </tr> <tr> <td>Architecture</td> <td><strong>Zen4 (2022)</strong></td> <td>Unknown</td> <td>Skylake (2017)</td> </tr> <tr> <td>Storage Type</td> <td><strong>Local NVMe (PCIe 4.0)</strong></td> <td>Network SSD</td> <td>Network SSD</td> </tr> <tr> <td>Memory</td> <td><strong>DDR5</strong></td> <td>DDR4</td> <td>DDR4</td> </tr> <tr> <td>YABS Runtime</td> <td><strong>8 min 53 sec</strong></td> <td>11 min 36 sec</td> <td>13 min 48 sec</td> </tr> </tbody> </table></div> <p>DigitalOcean's "CPU-Optimized" tier runs on Intel Xeon Platinum 8168 processors from 2017—that's 5-year-old silicon at 2025 prices. DanubeData runs AMD EPYC Genoa, the latest generation of datacenter processors.</p> <h2> Price-Performance: The Complete Picture </h2> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Metric</th> <th>DanubeData</th> <th>DO Premium AMD</th> <th>DO CPU-Optimized</th> </tr> </thead> <tbody> <tr> <td>Monthly Price</td> <td><strong>€8.99</strong></td> <td>$21</td> <td>$42</td> </tr> <tr> <td>Geekbench Single-Core</td> <td><strong>2,762</strong></td> <td>1,282</td> <td>1,098</td> </tr> <tr> <td>Points per Dollar*</td> <td><strong>~280</strong></td> <td>61</td> <td>26</td> </tr> </tbody> </table></div> <p>&lt;small&gt;*DanubeData converted at ~€1 = $1.05 for comparison&lt;/small&gt;</p> <p>DanubeData delivers:</p> <ul> <li> <strong>4.6x</strong> better price-performance than DO Premium AMD</li> <li> <strong>10.8x</strong> better price-performance than DO CPU-Optimized Intel</li> </ul> <h2> What This Means for Your Workloads </h2> <ul> <li> <strong>Databases:</strong> Up to 10x faster disk I/O means faster queries, quicker backups, and better transaction throughput</li> <li> <strong>Compilation &amp; Build:</strong> 2-3x CPU performance cuts build times dramatically</li> <li> <strong>Web Applications:</strong> Faster response times and higher concurrency</li> <li> <strong>Data Processing:</strong> Process datasets faster with modern Zen4 architecture</li> <li> <strong>Cost Efficiency:</strong> Get better performance while saving up to 79% on monthly costs</li> </ul> <h2> Reproducible Results </h2> <p>These benchmarks are fully reproducible. Spin up a DanubeData VPS and run the test yourself:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight shell"><code>curl <span class="nt">-sL</span> yabs.sh | bash </code></pre> </div> <p>Full Geekbench results:</p> <ul> <li><a href="https://browser.geekbench.com/v6/cpu/15429256" rel="noopener noreferrer">DanubeData DD Nano - Geekbench Result</a></li> <li><a href="https://browser.geekbench.com/v6/cpu/15429317" rel="noopener noreferrer">DigitalOcean Premium AMD - Geekbench Result</a></li> <li><a href="https://browser.geekbench.com/v6/cpu/15429932" rel="noopener noreferrer">DigitalOcean CPU-Optimized Intel - Geekbench Result</a></li> </ul> <h2> The Bottom Line </h2> <p>DigitalOcean's "CPU-Optimized" tier at $42/month runs on 2017-era Intel Xeon processors and gets outperformed by DanubeData's €8.99 entry-level VPS by a factor of 2.5x in CPU benchmarks and up to 10x in storage throughput.</p> <p>This isn't about minor optimizations—it's about running modern hardware. AMD EPYC Genoa (Zen4) with DDR5 and local NVMe storage simply operates in a different performance class than legacy infrastructure.</p> <p>Ready to experience the difference? <a href="https://danubedata.ro/register" rel="noopener noreferrer">Deploy your first VPS</a> and run the benchmarks yourself.</p> <h2> * Disclosure: This benchmark was conducted by DanubeData. We encourage you to run your own tests.* </h2> devops cloud vps benchmark