Live data from Hacker News

Benchmarking static website hosting providers

savjee.be

1–10 of 66 posts

Re: Benchmarking static website hosting providers

#2
CloudFlare's slowness is extremely surprising to me. From my house in Greece, I get a one millisecond ping to 1.1.1.1 (devices in my house have higher pings), so I was assuming that would carry over to their caches.

Too bad about that, looks like I'm going to have to rethink my setup of having them cache my site on their edge.

Re: Benchmarking static website hosting providers

#3
I would like to add Google App Engine CDN to the list. It is easy to use with Single Page Apps and also works with regular static websites. We have both use cases in the company I'm working at, and it works fine.

However custom domains (i.e. not appspot.com) have an increased latency that can be a huge issue in some regions, like Oceania.

Re: Benchmarking static website hosting providers

#4
Netlify results seem disappointing, I say that as a happy user of their service.

Netlify is an outlier in this comparison. For them, static site hosting is their core product, bread and butter if you will. To have such mediocre performance, as well as a relatively bad time to first byte is quite surprising.

Re: Benchmarking static website hosting providers

#5
A couple of months I did a similar benchmark of static hosts. Instead of relying on Pingdom I used Pulse by Turbobytes (which now seems dead) so the results were provided from end user networks all over the world.

https://www.pierbover.com/posts/static-hosting-benchmark-202...

Re: Benchmarking static website hosting providers

#6
A CDN is not the same as a hosting provider. Proximity to edge locations of the CDNs likely has something to do with latency delays. Also proximity of your hosting server location to Cloudflare's CDN.

I ran a similar benchmark for time to first bite with a Heroku site behind CloudFlare and since the Heroku edges location varied wildly, latency varied wildly as well. CloudFront in front of an S3 bucket in the same location would likely really fast. That said when you get hit by a bot attack using your registration form to spam QQ emails, you would be putting CloudFlare in front and that might be worth a benchmark as well.

One thing I would improve: benchmark of hosting provider speed and then benchmark the CDNs in front of specific hosting providers. If your hosting gives you a static IP to connect to your CDN, the speed will be less varied.

Re: Benchmarking static website hosting providers

#8
I put Cloudflare in front of my portfolio, which is hosted on Github Pages, so this made me do a bit of spit take. I always just assumed that GH Pages was a sort of “free perk for using Github” and that it must surely be hosted on “meh” infrastructure that would benefit from having a CDN in front. This article basically says that analysis is upside down.

I’ll need to see more benchmarks with a range of methodologies before I go in and redo how my portfolio is deployed, but this sure got me started searching.

Re: Benchmarking static website hosting providers

#9
post #2

CloudFlare's slowness is extremely surprising to me. From my house in Greece, I get a one millisecond ping to 1.1.1.1 (devices in my house have higher pings), so I was assuming that would carry over to their caches. Too bad about that, looks like I'm going to have to rethink my setup of having them cache my site on their edge.

1.1.1.1 doesn’t serve your websites. You’ll have to ping a hosted page.

Re: Benchmarking static website hosting providers

#10
post #9
post #2

CloudFlare's slowness is extremely surprising to me. From my house in Greece, I get a one millisecond ping to 1.1.1.1 (devices in my house have higher pings), so I was assuming that would carry over to their caches. Too bad about that, looks like I'm going to have to rethink my setup of having them cache my site on their edge.

1.1.1.1 doesn’t serve your websites. You’ll have to ping a hosted page.

He obviously understands that, hence the “I was assuming that would carry over to their caches.”
Post reply on HN