How to Read a VPS Specification Sheet: vCPU, RAM, and Bandwidth Explained

When you visit a VPS provider’s website, you are confronted with a grid of numbers: 1 vCPU, 2 GB RAM, 50 GB SSD, 2 TB transfer, 1 Gbps port. These numbers look simple, but they do not mean the same thing across providers. A “1 vCPU” from one host may perform half as well as a “1 vCPU” from another. “2 TB transfer” might be metered differently. This guide breaks down each spec on a typical VPS specification sheet, explains what the numbers actually mean in practice, and shows you how to compare apples to apples. To see how different providers stack up, browse our VPS comparison table for side-by-side specs.

vCPU: What You Are Actually Getting

A vCPU (virtual CPU) is a slice of a physical CPU core. How big that slice is depends entirely on the provider’s hypervisor and overselling ratio. The two main types are:

Shared vCPU (burst): Your VM borrows CPU time from a pool shared with other tenants. You get a baseline allocation (e.g., 10% of a core) and can burst above that when the host is not busy. Providers like DigitalOcean and Vultr use this model. Burst is great for spiky workloads but unreliable for sustained high CPU usage.
Dedicated vCPU: You get a guaranteed physical core or thread, no sharing. Providers like Hetzner and OVHcloud offer dedicated vCPU on some plans. This is essential for production databases, encoding, or any CPU-bound task that runs 24/7.

How to check: Look for the phrase “dedicated vCPU” or “guaranteed CPU” in the plan description. If it says “fair share” or “up to X cores,” assume it is shared. Run a sysbench CPU test after provisioning to see real performance:

sysbench cpu --cpu-max-prime=20000 run

Run this test multiple times at different hours. If the score varies by more than 20%, the CPU is shared and oversold.

CPU Architecture: x86 vs. ARM

Some providers now offer ARM-based VPS plans (e.g., AWS Graviton, Hetzner CAX). ARM instances are typically 20–40% cheaper than equivalent x86 instances. However, not all software is compiled for ARM. Docker images often have ARM64 variants, but proprietary binaries may not. If you run standard open-source software (Linux, Nginx, PostgreSQL, Redis, Python, Node.js), ARM is fine. If you need specific proprietary applications, verify ARM compatibility first.

RAM: Type, Speed, and Availability

RAM is the most straightforward spec — 2 GB is 2 GB. But there are caveats:

DDR4 vs. DDR5: DDR5 is faster and more power-efficient, but the difference is negligible for most VPS workloads. Do not pay extra for DDR5 on a budget VPS.
ECC vs. non-ECC: ECC (Error-Correcting Code) RAM detects and corrects memory corruption. It matters for databases and long-running servers. Most budget VPS providers use non-ECC RAM, which is fine for general use.
Swap vs. real RAM: Some budget providers advertise “2 GB RAM + 2 GB Swap” as 4 GB of memory. Swap is disk-based and roughly 1000x slower than RAM. Never count swap as usable memory. If a provider inflates their specs with swap, consider it a red flag.

Storage: SSD, NVMe, and IOPS

Storage is where specification sheets are most misleading. Here is what to look for:

Storage Type	Relative Speed	Best For	Typical Cost
SATA SSD	1x (baseline)	General web hosting, file storage	Lowest
NVMe SSD	3–5x	Databases, caching, high I/O apps	Moderate
NVMe (enterprise)	5–10x	Production databases, analytics	Highest

Many providers advertise “SSD storage” without specifying SATA or NVMe. If the plan is very cheap, assume SATA SSD. If it mentions “NVMe” explicitly, that is a real advantage. Run fio to test actual disk performance after provisioning:

fio --randrepeat=1 --ioengine=libaio --direct=1 --gtod_reduce=1 --name=test --bs=4k --iodepth=64 --size=1G --readwrite=randrw --rwmixread=75

Look for random read IOPS (input/output operations per second). A good NVMe should deliver 50,000+ IOPS. A SATA SSD might deliver 10,000–20,000 IOPS. If you get fewer than 5,000 IOPS, the storage is likely oversold or slow.

Bandwidth: Transfer vs. Port Speed

Bandwidth is the most commonly misunderstood spec. It has two components:

Data transfer (traffic cap): The total amount of data you can send and receive per month, measured in TB. A typical budget VPS gives 1–2 TB of transfer. This is a hard cap — exceed it and you either pay overage fees or get throttled.
Port speed (line rate): The maximum speed of your network connection, measured in Gbps. A 1 Gbps port means you can transfer data at up to 1 Gbps, but only until you hit your monthly transfer cap. At 1 Gbps, you can burn through 1 TB in about 2.5 hours of continuous full-speed transfer.

Real-world example: A VPS with 1 Gbps port and 2 TB transfer can sustain full speed for roughly 5 hours per month. After that, either the cap is hit or the provider throttles. For most web servers, this is fine — typical usage is 100–500 GB/month. But if you plan to run a file-sharing service, a video streaming site, or a backup server, calculate your expected transfer before buying.

Other Specs That Matter

DDoS Protection

Some budget VPS providers include basic DDoS mitigation (usually up to 10–20 Gbps). Others charge extra. If your VPS is publicly accessible, some protection is essential. A sustained DDoS attack with no mitigation can get your IP null-routed by the upstream provider.

Backup and Snapshot Support

Does the provider offer automated backups or manual snapshots? Some include one free snapshot; others charge per snapshot. On a budget VPS, you can set up your own backup script with rsync to an external storage provider, but built-in snapshots are more convenient for full-server recovery.

IP Addresses

Most budget VPS plans include one IPv4 address. Some providers charge extra for additional IPs or for IPv6 (though IPv6 should be free). If you need multiple IPs (e.g., for SSL certificates on different domains, or for outbound connections), check the price per additional IP.

How to Benchmark a VPS Before Buying

Most budget providers offer a money-back guarantee (7–30 days). Use that window to run benchmarks:

CPU: sysbench or 7-zip benchmark.
Disk: fio random read/write IOPS.
Network: iperf3 to a nearby test server, plus a speedtest-cli to check real-world bandwidth.
Latency: mtr to your target audience’s geographic region. High latency (>100 ms) will hurt web performance.

Run each benchmark three times at different times of day. If the results vary wildly, the provider is oversubscribed and you will experience inconsistent performance.

Putting It All Together

When comparing two VPS plans, ignore the headline specs and focus on the details:

Is the vCPU dedicated or shared?
Is the storage NVMe or SATA SSD?
Is the RAM real or inflated with swap?
What is the monthly transfer cap, and what happens when you exceed it?
Does the provider have a money-back guarantee so you can benchmark?

A VPS with 1 dedicated vCPU, 2 GB DDR5 RAM, 50 GB NVMe, and 2 TB transfer on a 1 Gbps port is a far better value than a plan with 2 shared vCPUs, 4 GB (including 2 GB swap), 50 GB SATA SSD, and “unmetered” bandwidth on a 100 Mbps port. The first one costs more on paper but delivers consistent performance. See our current VPS deals to find a plan with transparent specs and honest pricing.