Vultr vs Linode vs Digital Ocean vs Hetzner cloud network benchmark
I made some network tests to compare the 4 cloud providers (Vultr, Linode, Digital Ocean and Hetzner), in thse tests, all cloud instances were deployed in Frankfurt, Germany.
The tested instances are with ~similar CPU and Ram:
| Provider | CPU(s) | Ram | SSD capacity | Total traffic | Price |
| Vultr* | 2 | 4 GB ram | 128 GB Nvme | 3 TB | $25 /mo* |
| Linode | 2 | 4 GB ram | 80 GB SSD | 4 TB | $20 /mo |
| Digital Ocean | 2 | 4 GB ram | 80 GB SSD | 4 TB | $20 /mo |
| Hetzner | 2 | 16 GB ram | 40 GB SSD | 20 TB | $5.33 /mo |
A quick CPU test
First a quick CPU test, I downloaded a 1GB file from Leaseweb and I made the machines process this file with "cksum" and compared the time it took:
| Provider | command | result (lower is better) |
| Vultr | time cksum 1000mb.bin |
2.648s* |
| Linode | time cksum 1000mb.bin |
4.694s |
| Hetzner* | time cksum 1000mb.bin |
4.314s* |
| Digital Ocean | time cksum 1000mb.bin |
4.164s |
*this Vultr was a "high frequency" CPU version, there is also a $20 version
*also tested Hetzner 4 CPU version for ~$17 /mo (still cheaper than the rest), result: 0m2.809s.
Network upload test
These tests are made by downloading the same 1 GB file with wget, from one server to another, repeated the tests a few time, to make sure the speed is consistant.
Don't mix the directions :) I call this upload, when server uploads to other servers or to website visitors, this should be the important direction, the speed visitors gets to download from the server, is server upload speed:
| upload from | upload to | region | MB/s | latency | |
| Vultr | Linode | Germany | 383 | <1ms | |
| Vultr | Hetzner cloud | Germany | 220 | 3ms | |
| Vultr | Digital Ocean | Germany | 350 | 1ms | |
| Vultr | Digital Ocean | United States East | 30 | 87ms | |
| Vultr | Digital Ocean | United States West | 18 | 152ms | |
| Vultr | Digital Ocean | Singapore | 18 | 159ms |
| upload from | upload to | region | MB/s | latency | |
| Linode | Vultr | Germany | 170 | <1ms | |
| Linode | Hetzner cloud | Germany | 190 | 3ms | |
| Linode | Digital Ocean | Germany | 350 | 1ms | |
| Linode | Digital Ocean | United States East | 31 | 90ms | |
| Linode | Digital Ocean | United States West | 17 | 150ms | |
| Linode | Digital Ocean | Singapore | 17 | 166ms |
| upload from | upload to | region | MB/s | latency | |
| Hetzner | Vultr | Germany | 265 | 3ms | |
| Hetzner | Linode cloud | Germany | 328 | 3ms | |
| Hetzner | Digital Ocean | Germany | 300 | 3ms | |
| Hetzner | Digital Ocean | United States East | 30 | 80ms | |
| Hetzner | Digital Ocean | United States West | 17 | 165ms | |
| Hetzner | Digital Ocean | Singapore | 16 | 170ms |
| upload from | upload to | region | MB/s | latency | |
| Digital Ocean | Vultr | Germany | 190 | 1ms | |
| Digital Ocean | Linode cloud | Germany | 240 | 1ms | |
| Digital Ocean | Hetzner | Germany | 160 | 3ms | |
| Digital Ocean | Digital Ocean | United States East | 32 | 83ms | |
| Digital Ocean | Digital Ocean | United States West | 17 | 158ms | |
| Digital Ocean | Digital Ocean | Singapore | 17 | 158ms |
Download test
I didn't do a separate table, but you can reversely look in the above upload test and see the server you are interested.
My conclusion
What I understand from the tests is that they are all great cloud providers;
Generally faster and cheaper than some competitors I tested, I also use OVH for example.
If you need more than 3-4 TB traffic you might look more into Hetzner, also if you want to get decent performance for as low as $5-$10 but without much storage Hetzner would fit.
In my case now I want it be be as responsive as possible so this particular situation I will go with the Vultr "High Frequency" version. Hetzner and Linode also have a "Dedicated CPU" improved performance instances but they are a bit more expensive.
What I also noticed is that Hetzner are a bit more strict when it comes to abuse or external complaints, I had 2-3 situatons when users uploaded inapropriate content on my site(s) and Hetzner disconnected them in a few hours and re-activated the server after I removed the content. In another situation whwere I hosted a mobile apps directory, Hetzner simply asked me to move the site out of their network, without any room for discussion. So yes, they are a bit more sensitive; this also means that their IPs/network has a better reputation, e.g spam scores and such.