TranscomAll articlesHome

How Much RAM, CPU and Storage Does Your VPS Actually Need? A Practical Sizing Guide

Updated 16 November 2026

Choosing a VPS should be simple, but the spec sheets make it feel like guesswork. Pick too small and your site slows down or crashes under load. Pick too large and you pay every month for capacity you never use.

This guide explains what each resource actually does and gives realistic starting points for common workloads. It also shows how to check whether you got it right.

Start with the workload, not the spec sheet

Before comparing plans, write down what the server will run. Be specific. "A website" is vague. "WordPress with MariaDB, Nginx, PHP-FPM and a nightly backup script" is something you can size.

For each component, ask three questions:

RAM: usually the first limit you hit

For most web and application servers, memory runs out before anything else. When Linux runs out of RAM it starts swapping to disk, which is far slower. If that is not enough, the kernel's out-of-memory killer terminates processes, and your database is often the first casualty.

As a rough guide:

Windows Server needs considerably more memory than a minimal Linux install. Budget accordingly if you plan to run it.

vCPU: when more cores really help

CPU matters most when lots of things happen at the same moment. More vCPUs let the server handle more simultaneous requests, worker processes or build jobs without them queuing.

You will feel the benefit of extra cores with:

A quiet blog or a single background service will barely use one core. Doubling the vCPU count will not make a memory-starved server faster, so diagnose before you upgrade.

Storage: size and speed

Disk space is easy to estimate. Add up your operating system, applications and current data, then leave generous headroom for growth, logs and any local copies of backups. Running a disk close to full causes failed writes, broken updates and corrupted databases, so aim to stay well below capacity.

Speed matters too, especially for databases. NVMe storage handles many small random reads and writes far better than older disk types, which keeps queries responsive under load.

Some workloads need far more space than memory or CPU. Examples include file archives, media libraries, backup targets and large log stores. For these, a high-storage plan is better value than buying a bigger all-round server just to get the disk.

The Transcom VPS range at a glance

Every Transcom VPS runs on KVM virtualisation with dedicated RAM, CPU and disk. Every plan includes NVMe storage, unlimited bandwidth, IPv4 and IPv6, full root access, free backups and DDoS protection. Prices are per month, plus VAT.

PlanRAMvCPUNVMePriceSuited to
HA-512512 MB120 GB£2.00Lightweight always-on services
HA-1G1 GB150 GB£4.00Small sites and single services
HA-2G2 GB280 GB£7.00General-purpose hosting
HA-4G4 GB280 GB£13.00Busier sites and application stacks
SS-4G4 GB480 GB£15.00Compute-heavy and concurrent workloads
SS-NOMINET4 GB680 GB£18.00Nominet / registrar workloads
HA-8G8 GB3100 GB£24.00Databases and multi-service stacks
HV-8G8 GB31 TB£28.00Storage-heavy workloads
HA-16G16 GB4200 GB£45.00Demanding production applications
HA-32G32 GB6300 GB£85.00The most demanding production workloads

Note the side-by-side choices. If you need 4 GB of RAM, the SS-4G trades a little extra cost for double the cores of the HA-4G. If you need 8 GB, the HV-8G swaps the HA-8G's 100 GB of disk for a full terabyte.

How to tell if you sized it wrong

Once your server is live, a few standard Linux commands tell you whether it fits:

  1. free -h shows memory use. Look at the "available" figure. If it stays low and swap use keeps climbing, you need more RAM.
  2. uptime or top shows the load average. A load consistently above your vCPU count means work is queuing for the processor.
  3. dmesg | grep -i oom reveals whether the out-of-memory killer has been terminating processes.
  4. df -h shows disk usage per filesystem. Act well before any filesystem gets near full.
  5. vmstat 5 or iostat highlights processes waiting on disk, which points to a storage bottleneck rather than CPU.

Check these at your busiest times, not at 3am. A server that looks idle overnight may be struggling at lunchtime.

Start small, then scale with evidence

The best approach is usually to choose a sensible starting tier, measure real usage for a few weeks, then adjust. Guessing high "just in case" is the most common way to overspend.

Transcom VPS plans are month-to-month with no contract, so you are not locked into a size that turns out to be wrong. Free backups are included, and your firewall, console, reboot and reinstall controls all sit in your own panel, which makes moving to a better-fitting plan less daunting.

Ready to choose?

If you are running a typical website or a small application stack and are not sure where to begin, the HA-2G KVM VPS is a sensible general-purpose starting point. It comes with 2 GB of dedicated RAM, 2 vCPU and 80 GB of NVMe storage for £7.00 + VAT per month. It is UK-hosted, online within minutes and has no contract, so you can measure your real usage and change plan if your workload needs it.