VPS Linode vs Scaleway vs DigitalOcean

[I A 1]

Recently we migrated our forum from a dedicated server to Linode and saved almost 75% overall hosting cost. Our ad revenue is quite low and sometimes I feel we must try cutting costs even more.

Scaleway offers very cheap hosting, starting at 2.99 Euro per month, with 2GB RAM and 50GB SSD.

I made a comparison check between the servers and found that Linode runs on Intel Xeon E5-2680 and Scaleway runs on Intel Atom C2550 or C2750 depending on the package. Here's a comparison page for them: http://ark.intel.com/compare/77982,77987,64583

Has anyone used scaleway for hosting a medium sized forum? How was the performance?

 

[Dan18]

You better stick with Linode.

 

[Nominance]

I have used Linode in the past and they are pretty good. I would recommend them

 

[eva2000]

Linode runs on Intel Xeon E5-2680 and Scaleway runs on Intel Atom C2550 or C2750

Huge diff in performance in favour of Linode so best to stick with Linode. Though depends on your server usage requirements.

 

[VerniceArielle]

My husband is suggesting Linode. He has a good experience using it. According to him, if a developer needs to get the most out of their server then you can rely on it. Overall, most reviews will say that Linode delivers a very good cloud hosting.

 

[I A 1]

So today I wanted to test the new droplets offered by DigitalOcean and compare it with my existing VPS on Linode. On Linode, I have 4 GB RAM, 2 CPU cores, 48 GB SSD at $20 per month, so I select the similar one DO with 4 GB, 2 CPU and 80 GB SSD. After the complete setup, I used Plesk migrator tool to copy the websites, and assigned them free tlds from freenom.com

Now I have two servers running the same websites with different names. One is live and the other has no traffic.

Linode VPS is running on Intel E5-2680 v3 and the DigitalOcean droplet is running on E5-2650 v4. Compare them here.
Testing both of these using Google's PageSpeed Insights, the live site on Linode has one number higher optimization compared to DO.

And using webpagetest.org, I get the following for Linode:


and for DigitalOcean, for the same thread.

The only difference between them is, the site on Linode has https certificate, but on DigitalOcean, I could not issue the certificate and simply used http for the thread link.

DigitalOcean is obviously faster, but could it be because this is just a test forum with no live traffic? What other test do you suggest before I actually decide to migrate?

 

[djbaxter]

I get a heck of a lot of blocked spam and brute force attempts coming from Digital Ocean servers, if that concerns you at all.

Be wary of bad neighborhoods.

 

[mysiteguy]

Suggestion: create a sub-domain, assign it's DNS to the Digital Ocean droplet, install a free Let's Encrypt certificate and retest. You need to do an apples to apples comparison since the HTTPS handshake has some overhead which effects time to first byte.

 

[Pete]

So today I wanted to test the new droplets offered by DigitalOcean and compare it with my existing VPS on Linode. On Linode, I have 4 GB RAM, 2 CPU cores, 48 GB SSD at $20 per month, so I select the similar one DO with 4 GB, 2 CPU and 80 GB SSD. After the complete setup, I used Plesk migrator tool to copy the websites, and assigned them free tlds from freenom.com

Now I have two servers running the same websites with different names. One is live and the other has no traffic.

Linode VPS is running on Intel E5-2680 v3 and the DigitalOcean droplet is running on E5-2650 v4. Compare them here.
Testing both of these using Google's PageSpeed Insights, the live site on Linode has one number higher optimization compared to DO.

And using webpagetest.org, I get the following for Linode:
View attachment 50743

and for DigitalOcean, for the same thread.
View attachment 50744
The only difference between them is, the site on Linode has https certificate, but on DigitalOcean, I could not issue the certificate and simply used http for the thread link.

DigitalOcean is obviously faster, but could it be because this is just a test forum with no live traffic? What other test do you suggest before I actually decide to migrate?

Do the test on both without HTTPS and on both with HTTPS. The actual handshake plus the encryption makes a difference.

Also, in terms of things that might sway your decision, what support do you need? I once broke my Linode so hard it wouldn't boot (because I didn't know what I was doing), and they fixed it for me in about 15 minutes. I have no experience of DO though.

 

[ENF]

Voting for Linode, heavily.

Here's a test of our main server at Linode. (We're Asia/Japan based, but have footprints in each of the Linode DC's)



(We don't use an outside CDN and this production server is running under ~95% load at the time of this test above, cpu graph below)



For the cost vs. performance aspect, we are very pleased to have moved all of our infrastructure into Linode.

 

[I A 1]

I have tested using https on both and the time to first byte is better on DO, but still not as good as shown above. How do we improve it? My forum is running vB 4.2.4 and has around 1M posts and 3-4k daily visitors. Do you suggest a better VPS?

 

[Paul M]

I have never hear of Scaleway.

Linode are very good, and DO are also good, but I actually dont use either for my main server - thats currently on Vultr.

 

[ENF]

I have tested using https on both and the time to first byte is better on DO, but still not as good as shown above. How do we improve it? My forum is running vB 4.2.4 and has around 1M posts and 3-4k daily visitors. Do you suggest a better VPS?

I should have added specifics on our usage and setup...

> XF 1.5.18 (Elasticsearch, PHP 7.0.x, HTTPS) ~300 constant online with bursts of ~600-800 at peak times. (1 site only)
> CentOS 7.x / Centminmod Base Platform w/Tuning adjustments for server sizes and traffic amount.
> Linode 12288 (6 CPU Cores, 12GB RAM) for Primary Servers [One of those is shown in the example in my previous post]
> Linode 8192 (4 CPU Cores, 8GB RAM) for Backup/Testing Nodes

 

[eva2000]

Linode has best performance out of the 3 titled hosts but pagespeed is related to geographic location of your test servers as you can see below in my California vs Dulles WPT tests.

I should have added specifics on our usage and setup...

> XF 1.5.18 (Elasticsearch, PHP 7.0.x, HTTPS) ~300 constant online with bursts of ~600-800 at peak times. (1 site only)
> CentOS 7.x / Centminmod Base Platform w/Tuning adjustments for server sizes and traffic amount.
> Linode 12288 (6 CPU Cores, 12GB RAM) for Primary Servers [One of those is shown in the example in my previous post]
> Linode 8192 (4 CPU Cores, 8GB RAM) for Backup/Testing Nodes

Nice to see my Centmin Mod LEMP stack speeding up your site

Running my Centmin Mod forums https://community.centminmod.com/ on single Linode 8GB VPS in Fremont, California with latest Centmin Mod 123.09beta01 with Nginx 1.13.12 + ngx_pagespeed, PHP-FPM 7.1.17

Running with HTTP/2 based HTTPS with XF 1.5.19 with Elasticsearch

WPT California




WPT Dulles, VA