Best VPS for databases
Databases live and die by disk I/O and memory. This ranking weights random disk IOPS and memory bandwidth the most, then CPU.
Developer cloud · SSD droplets · global datacenters
| # | Provider | Score | Relative | Tests |
|---|---|---|---|---|
| 1 |
Ghostnet
|
87.2 |
|
1 test |
| 2 |
Vultr
|
84.3 |
|
5 tests |
| 3 |
Rj Network
|
73.1 |
|
1 test |
| 4 |
Trabia
|
72.5 |
|
1 test |
| 5 |
M247 Europe
|
71.2 |
|
2 tests |
| 6 |
Waicore
|
70.8 |
|
1 test |
| 7 |
Serva One
|
67.1 |
|
10 tests |
| 8 |
Tihost
|
66.4 |
|
1 test |
| 9 | T Timeweb | 66.3 |
|
2 tests |
| 10 |
Netgrid Host
|
55.8 |
|
13 tests |
| 11 |
ITLDC
|
47.9 |
|
3 tests |
| 12 |
Digitalocean
|
46.7 |
|
7 tests |
| 13 |
Edis
|
31.0 |
|
1 test |
A database is bound by disk and memory, not the network. So the VPS for databases (MySQL, PostgreSQL, MongoDB) ranking values random disk IOPS and memory bandwidth most, then CPU. The best database VPS right now is Ghostnet.
VPS for databases: why the disk outweighs gigahertz
The database score comes from disk (40%), memory (35%) and CPU (25%). The network takes no part: a database almost always sits next to the application and could not care less about milliseconds to Cloudflare.
Disk leads for a reason. Real queries are random 4–8 KB operations, not the pretty sequential reads used in marketing. A server advertising «5 GB/s» can collapse on random writes — and random writes are exactly what we measure.
Memory comes second: the more of the working set fits in RAM, the less the database touches the disk at all. Past a certain size that stops being an optimisation and becomes the difference between «fast» and «not happening».
The selection holds 13 hosts, led by Ghostnet at 87.2. The gap between first and last is 56.2 points. That is a lot: the choice here genuinely matters rather than being a matter of taste.