How Much RAM Does a Small Website Need

There is no single RAM figure that fits every small website. A site serving cached pages uses memory differently from a shop calculating carts or an application processing uploads. Start with the software you run and how it behaves under realistic use.

RAM holds data that running processes need quickly: the operating system, web server, application workers, database, and background jobs. If the database runs on a separate service, its memory is not part of your web server's budget.

Count the work happening at once

Monthly visitor totals hide short busy periods. A newsletter can bring many visitors at once, and a scheduled job can run while people browse. List the processes on your server and look for tasks that overlap: page requests, admin screens, imports, image processing, backups, and deployments.

Build an estimate from measurements

A first estimate is: memory for the operating system and always-running services, plus active application workers, plus room for background jobs.

Here is a rough paper calculation, just to show the math:

• Operating system and services: 500 MiB
• Four active workers at 120 MiB each: 480 MiB
• A scheduled task running at the same time: 200 MiB
• Subtotal: 1,180 MiB before headroom

Those inputs must come from your own server before you use them to buy anything.

What each step up costs

Once you know roughly how much memory you need, the next question is what the next size costs. These are self-managed VPS plans from our price tables:

Both providers' prices above are monthly. Vultr bills hourly and DigitalOcean by the second, both up to that monthly price. Both let you move to a larger plan later (Vultr does not support shrinking a server, DigitalOcean can't shrink a disk once it has been enlarged, and a resize means some downtime), so starting at a measured size and moving up when the numbers show pressure is usually cheaper than buying extra headroom on day one. Full details are on our Vultr and DigitalOcean plan pages.

Read the right signals

On Linux, low "free" memory does not mean the server is struggling; the system uses spare memory for caches. The MemAvailable value estimates memory available for new programs without swapping. [1]

Watch for out-of-memory events, failed jobs, slower responses, and constant swap use. A slowdown is not always a memory problem: CPU, disk, network, or database design may be the cause.

If you use PHP, memory_limit caps what one script may use. It is not the server's total RAM, and raising it does not add memory. [2]

Test before you commit

Test ordinary browsing, admin tasks, and the heaviest routine jobs in a test environment, and record memory, response time, and errors together. Before upgrading, check for unnecessary services, too many workers, and expensive jobs, and watch the memory graphs over time. [3]

Sources

Checked 2026-10-14. The worker numbers above are a hypothetical example to show how the math works. Prices are from official pricing pages on the dates shown.

• [1] https://docs.kernel.org/filesystems/proc.html
• [2] https://www.php.net/manual/en/ini.core.php
• [3] https://docs.digitalocean.com/products/droplets/concepts/choosing-a-plan/