What is a VDS and how does it differ from a VPS?
Most providers use the terms VDS and VPS almost interchangeably, yet the difference between them affects both your bill and your server's performance. The distinction is not in the name — it is in how resources are shared.
The short answer
A VPS (Virtual Private Server) is created by virtualising a physical server, and its resources are shared. You may have been allocated 4 vCPU, but that does not mean those four cores are yours at every moment; when other customers on the same physical machine get busy, your work waits.
A VDS (Virtual Dedicated Server) uses the same virtualisation technology, but resources are reserved for you. If you bought 4 vCPU, that capacity is not shared with another customer.
That is the technical distinction. Because the industry uses both terms loosely, however, you should judge the provider's commitment rather than the label.
The real issue: overselling
Overselling is when a provider sells more resources than the physical machine actually has. A server with 128 GB of RAM might have 300 GB sold across its customers, on the assumption that they will not all use their full allocation at once.
That assumption usually holds — until it does not. Typical symptoms on an oversold machine:
- The server feels normal during the day and noticeably slow in the evening
topshows the CPU as idle, yet processes still wait- I/O wait rising for no apparent reason
- The same code running twice as slowly a week later on the same server
None of this appears as a measurable fault on your server, because the problem is not in your virtual machine — it is in the physical machine underneath it.
Comparison
| VPS (shared) | VDS (reserved) | |
|---|---|---|
| vCPU | Shared, neighbour load affects you | Reserved, capacity guaranteed |
| RAM | Usually reserved, sometimes inflated | Reserved |
| Disk I/O | Shared, the most volatile factor | Reserved or rate-limited |
| Price | Lower | Higher |
| Suited to | Static sites, test environments, low-traffic panels | Game servers, databases, production applications |
Four questions to ask before buying
Questions whose answers you can actually verify, instead of reading marketing copy:
- "Do you oversell?" If you cannot get a clear "no", they do.
- "Is my vCPU guaranteed or shared?" If the answer is "guaranteed", you can verify it by running a single-core benchmark at peak hours.
- "Is the disk NVMe or SATA SSD?" If you will run a database, this difference is measured in multiples rather than percentages.
- "Where is the server physically located?" If you serve users in Türkiye, an overseas location means 40-60 ms of added latency — something players notice directly on a game server.
Which should you choose?
The decision rule is simple: if your workload is latency-sensitive, choose VDS; if not, VPS.
Latency-sensitive workloads: game servers (FiveM, Minecraft, Rust and similar), real-time applications, sites with an active database workload, voice servers.
Workloads that are not: static sites, small services running scheduled jobs, development and test environments.
We have written separately about how to size resources by workload type: How many vCPUs and how much RAM does a server need?
How this works at Datafex
On Datafex virtual servers, resources are reserved and we do not oversell. Every plan uses NVMe SSD storage, and servers are hosted in the DGN Data Centre in Bursa, with 100 Gbps domestic and 10 Tbps international transit capacity. Fexwall DDoS protection at L3/L4 is included free on every plan.
If what you need is a game server, we have compiled RAM, vCPU and storage guidance per game by player count: game server plans.