Back to the journal

VPS RAM and CPU: How Much Do You Need?

For VPS RAM and CPU, a small static site or lightweight bot may start with 1–2 GB of RAM and 1 vCPU. A web application with a local database may need 4 GB and 2 vCPUs or more. These are starting estimates: simultaneous work, database size, and peak usage determine the right plan.

By Drew F. · Last updated: November 17, 2026

Start with the work your server will do

List every service that will run at the same time: the operating system, web server, application workers, database, background jobs, monitoring, and deployment tools. A container is a packaging boundary, not free capacity; its processes still use the VPS's memory and CPU.

Next, identify the busiest period. Ten thousand visitors spread across a day can be easier to serve than a smaller group making expensive requests at once. Scheduled backups, imports and builds can overlap with traffic, too. Size for that overlap, not just an idle dashboard.

How much RAM does a VPS need?

RAM holds running programs and data they need quickly. If the working set cannot fit, the system may reclaim caches, swap memory to disk, slow down or terminate processes. Disk capacity is separate: a 20 GB database does not automatically require 20 GB of RAM, but a large active working set benefits from more memory.

Use these rough starting points for a Linux server, then measure under representative load. They are planning estimates, not guarantees or Ingen plan specifications.

Workload Starting RAM Starting CPU Assumptions
Static site, small bot or light monitoring 1–2 GB 1 vCPU Few processes and low concurrency
Small dynamic site with a local database 2–4 GB 1–2 vCPUs Modest traffic and controlled worker counts
Small app, database and background worker 4–8 GB 2–4 vCPUs Multiple services and occasional concurrent jobs
Several apps or a busier database 8 GB or more 4 vCPUs or more Actual demand varies widely; load test

Check your software's published minimums and recommended requirements first. Windows, control panels, large plugins, search indexes, and game servers can require substantially different amounts. A minimum that boots is not necessarily enough for a production workload.

Work through an estimate

Suppose a small app's measured peak is 1.2 GB, its database uses 1 GB, and the operating system and other services need 0.6 GB. That is 2.8 GB before extra room for caching, traffic spikes and maintenance. A 4 GB server is a reasonable candidate to test, not a promise it will fit. If a build or backup overlaps with traffic, include its peak too. Avoid adding per-process memory numbers blindly: shared memory can otherwise be counted more than once.

How many vCPUs do you need?

A vCPU is a virtual processing unit assigned to your server. Its sustained performance depends on the underlying hardware and how the provider allocates CPU time; two vCPUs on different plans are not necessarily equivalent. Ask whether CPU capacity is shared, dedicated or subject to burst limits before comparing numbers.

One vCPU can be a reasonable starting point for a light workload. Two or more can help when several requests or jobs need CPU at once. More vCPUs do not necessarily speed up a single-threaded task, and they will not fix a database waiting on slow storage or an application blocked on a remote API.

Look at the work itself: image processing, compiling and encryption can consume CPU; a cached static response often needs much less. Measure response times and throughput during representative peaks rather than choosing cores from monthly visitor counts.

Once you have a candidate size, review Ingen Cloud's VPS options and confirm the current plan details and upgrade process before ordering.

Check memory and CPU after launch

Monitor during normal use, the busiest expected traffic, and maintenance jobs. Record application response time alongside resource use; a server that looks quiet while users wait may be blocked on a database, disk or network dependency.

The following Linux commands are examples for Ubuntu Server 24.04. They are read-only, but results vary with installed tools and permissions.

Run free -h to view memory and swap totals in human-readable units:

free -h

Look for the available column in the Mem: row. It estimates memory available for new applications without swapping; free alone omits memory that can be reclaimed from cache. The Linux kernel documentation explains how MemAvailable is estimated. Swap in use is not by itself proof of a current problem: check whether usage is growing and whether the application is slowing down.

Run top to inspect live CPU and memory use by process:

top

The display updates repeatedly; press q to quit. Compare active processes with slow periods. Persistent CPU saturation alongside a growing request queue can indicate CPU pressure. Load average alone is not a CPU utilization percentage: it also includes tasks in uninterruptible sleep, often while waiting for I/O. See the Linux proc manual.

If memory is tight, inspect worker limits, caches and database settings before upgrading. If CPU is busy, identify the expensive task and consider caching, scheduling jobs away from peaks or adding CPU. If neither is constrained, check disk latency, network dependencies and application logs. A single snapshot is a clue, not a capacity plan.

When is a VPS the wrong choice?

A VPS requires someone to operate the OS and application stack. If you do not want that responsibility, a managed application platform or hosting service may fit better. If the workload needs guaranteed sustained compute, a GPU, or resources beyond the available VPS sizes, evaluate a suitable dedicated or specialized service. Confirm the provider's actual guarantees and limits in either case.

Plan for growth

Start with enough capacity for the measured or estimated peak and a sensible margin. Before launch, check whether resizing changes the price, requires downtime or affects storage, and keep a tested backup. After launch, review usage during representative peaks and after major feature or traffic changes. The server selection guide provides a broader decision framework; the VPS getting-started guide covers the next stage, and the VPS security checklist is useful when preparing the server.

Frequently asked questions

Is 2 GB of RAM enough for a VPS?

It can be enough for a small Linux service or lightly used site, depending on the software and concurrent work. A local database, multiple workers or memory-heavy plugins can make it too small. Check the application's requirements and measure peak available memory.

Should I upgrade RAM or CPU first?

Upgrade the resource that constrains the workload. Low available memory with active swapping and slowdowns suggests memory pressure; sustained CPU saturation during slow requests suggests CPU pressure. Check storage and external dependencies as well.

Can I add RAM and CPU later?

That depends on the provider and plan. Confirm available upgrade paths, downtime, pricing and any storage restrictions before relying on a later resize.