It was tested: Google's C3D as well as C2 as well as N2 virtual machine for customers (r)

Sep 26, 2024
An illustration representing a bar chart and the words, 'C3D speed test.'

-sidebar-toc> -language-notice>

Below, we will review the data from testing conducted on real WordPress websites.

What's new with Google's C3D machine?

  1. The X4 is the Fourth-generation AMD EPYC CPU (formerly codenamed "Genoa") which can be operating at 3.7 7 GHz. It can support up to 360 virtual CPUs as well as over 2,880GB of DDR5 memory.
  2. A Infrastructure Processing Unit (IPU) which improves network performance and data I/O, while freeing the CPU from those tasks.

WordPress websites that are hosted by them can make use of this capability because they run in separate containers from other websites and comes with all required software like Linux, NGINX, PHP and MariaDB.

Performance comparisons between VMs

To test the performance of C3D's VM as compared to other types that use computers, we built three identical WordPress (v6.5) sites that were built using the C3D machine. One of them was on one of the C2 machines, the final one on N2 computer that is available at a couple various data centers. (Although the utilization of N2 computers isn't heavy in other hosting providers, a lot of hosting providers use them, and we've included this study to illustrate the effect it will have on the speed your site is switching from .)

It was the WordPress hosting environment that was shared by the three websites:

  • WordPress version 6.5
  • Version 8.2 of PHP. 8.2
  • Ubuntu version 20.04.6
  • NGINX web server version 1.25.2
  • MariaDB Version 15.1

In order to simulate the load of our test sites, we utilized the Apache ab HTTP Server benchmarking application it has the capability of simulating many simultaneous users. It can create many different websites.

The demand for content that is not cached

In the absence of cached content WordPress must query the database to create the desired page per request. This isn't efficient, however it's typically required if the content is exclusive to a specific user such as what's in an online shopping cart.

For each of the three tests our test sites included the default shopping cart view generated a 235KB HTML web page.

The test protocol of the Apache ab test protocol looked something like the following:

  • Dimension of page is 235 KB
  • Simulated concurrent users: 50
  • Run time: 60 seconds

The result (successful requests in a second):

  • C3D: 207.72
  • C2: 141.47
  • N2: 88.93

Summary Conclusion C3D Server: The C3D server provided on average 46.8 percent more cached pages than the twin site on the C2.

Chart showing results of uncached page-request testing for N2, C2, and C3D virtual machines.
Test results on speed and performance for cache-free (cache-bypass) web pages.

Demanding cached content

By enabling caching, our WordPress websites can provide pages, without the need to use PHP employees or query the database. Indeed, thanks to the cache in memory of NGINX, certain created content does not even need to be read-only stored.

Our content test cachable consisted of the same blog post for each site which weighed in at 114KB. Thus our testing protocol with Apache ab test protocol looked like this:

  • Size for the webpage of 114KB
  • Simulated concurrent users 50
  • Run time: 60 seconds

The result (successful requests in a second):

  • C3D: 19,722.58
  • C2: 13,043.27
  • N2: 7,861.23

Summary Summary: The C3D VM thanks to its upgraded I/O stood out here in the sense that it moved content faster than all other machines. The C3D website produced 51.2 percent more page cached than C2's machine.

Chart showing results of cached page-request testing for N2, C2, and C3D virtual machines.
The speed tests for requests to cached pages.

One of the best examples of the raw processing power

Image manipulation was used to assess the performance of our VMs but it wasn't directly linked to the delivery of web pages, however the process of resizing uploaded files and making duplicates with different sizes is a standard procedure on many WordPress websites.

We applied the ImageMagick extension within PHP to shrink the size of a 35MB JPEG image down to around 29KB (from 7,362 4.702 x 4,702 pixels down to 408x640 pixels) using that software's ResizeImage() function and the Bessel filter.

The resultant average processing time:

  • C3D 1.484 seconds
  • C2 2.090 seconds
  • 2: 2.305 seconds

Summary Resizing the image was fairly fast for all the testing platforms, the C3D machine accomplished the task about 30 percent quicker than the C2 machine.

A chart showing the speeds at which C3D, C2, and N2 VMs resized a large image.
Results of tests for speed of image processing.

The C3Ds aren't in the data centre? That's fine!

C3Ds provide a significant advantage for the owners of WordPress websites. Particularly, if your site creates content that cannot be stored, consider if those advantages make it worthwhile to house your content in one of the Data Centers with the most recent technology.

Edge caching could be the best solution for speeding the performance of those who need to locate their website in a location with faster VMs cannot be found.

Edge-caching content is distributed across Cloudflare's global data centers the response time for web pages on C3D, N2, and C3D, C2, and N2 test machines were practically similar when they were compared with time to"the first bytes" (TTFB) in various other locations.

A chart showing time-to-first-byte on C3D, C2, and N2 VMs when all are supported by edge caching.
The time-to-firstbyte outcome with Edge Caching activated.

Even when using Edge Caching, average response speed over time can be slightly faster on C3D machines since sites hosted on them can be full and refresh caches quicker.

What are the best places to look for C3D machines?

Presently, C3D machines are available for the following 8

Google Cloud data centers:

  1. Jurong West, Singapore (asia-southeast1)
  2. Sydney, Australia (australia-southeast1)
  3. St. Ghislain, Belgium (europe-west1)
  4. Frankfurt, Germany (europe-west3)
  5. Eemshaven, Netherlands (europe-west4)
  6. Council Bluffs, Iowa, USA (us-central1)
  7. Moncks Corner, South Carolina, USA (us-east1)
  8. Ashburn, Virginia, USA (us-east4)

In My Dashboard My Dashboard, regions with the C3D devices enabled are identified with the words "Boosted in the data center drop-down menu" when creating a website. WordPress website:

A screenshot of the My dashboard showing the dialog used to select a data center.
Data centers with C3D machines are flagged as increased within My.

For now, Google's compute-optimized C2 VMs are the most efficient at these data centers.

  1. Changhua County, Taiwan (asia-east1)
  2. Hong Kong (asia-east2)
  3. Tokyo, Japan (asia-northeast1)
  4. Osaka, Japan (asia-northeast2)
  5. Seoul, South Korea (asia-northeast3)
  6. Mumbai, India (asia-south1)
  7. Delhi, India (asia-south2)
  8. Jakarta, Indonesia (asia-southeast2)
  9. Sydney, Australia (australia-southeast1)
  10. Melbourne, Australia (australia-southeast2)
  11. London, United Kingdom (europe-west2)
  12. Frankfurt, Germany (europe-west3)
  13. Zurich, Switzerland (europe-west6)
  14. Montreal, Canada (northamerica-northeast1)
  15. Toronto, Canada (northamerica-northeast2)
  16. Sao Paulo, Brazil (southamerica-east1)
  17. Columbus, Ohio, USA (us-east5)
  18. The Dalles, Oregon, USA (us-west1)
  19. Los Angeles, California, USA (us-west2)
  20. Salt Lake City, Utah, USA (us-west3)
  21. Las Vegas, Nevada, USA (us-west4)

Summary

Our testing suggested that moving to Google's latest, more efficient C3D virtual machines could be beneficial to the webmasters of several websites, especially those who have sites with data that's not cache-able.

In comparison to only our C2 machines -previously the fastest machines used to host sites in Google's data centers C3Ds were faster. C3Ds demonstrated:

  • A performance improvement of almost 47% on pages that are not cached
  • An increase of around 52% in the cached page's request
  • A 30percent improvement in processing time for a task like resizing images

Steve Bonisteel

Steve Bonisteel is a Technical Editor for the site. He started his career in writing as a reporter for newspapers, who chased ambulances and fire trucks. He's been reporting on developments in technology on the Internet since the 1990s.

Article was posted on here