Why we write these
Most hosting comparisons conclude that the author's product wins. That's not useful when you're deciding at 11pm whether to move a service. These pages lay out the mechanics: what each option costs you in money, time, latency and operational burden, and which workloads tip the balance.
We sell one thing: self-managed VMware virtual machines in London with NVMe-backed storage, a 1 Gbps unmetered port and one IPv4, billed monthly or annually in US dollars. Keep that in mind as you read. Where that product doesn't fit, the pages say so.
VPS vs serverless
Functions-as-a-service bill per invocation and scale to zero. A VPS bills a flat rate and runs all the time. The crossover depends on traffic shape: spiky, low-volume event handlers are cheap on functions; steady request load, long-lived connections, background workers and anything needing a local disk tend to be cheaper and simpler on a server. Cold starts, execution time limits and per-request pricing are the main serverless costs that don't show up on the first invoice.
VPS vs dedicated
A dedicated server gives you every core, every gigabyte and the whole disk, with no neighbours. A VPS slices a host into isolated virtual machines and sells you a slice. Dedicated hardware wins on sustained heavy CPU, very large memory footprints and workloads that need specific hardware. VPSs win on time to provision, price at small sizes, and the ability to rebuild from an image in minutes. We only sell the VPS side.
Virtual machines vs containers
These are complementary rather than competing. A hypervisor virtualises hardware, so each server has its own kernel; containers share the host kernel and isolate processes with namespaces and cgroups. The page covers isolation strength, kernel control, density, and the very common pattern of running containers inside a virtual machine, which is what most of our customers do.
NVMe vs SATA SSD
Both are flash. The difference is the interface: SATA was designed for spinning disks and carries a single command queue, while NVMe talks to the CPU over PCIe with many parallel queues. That shows up as lower latency and far better behaviour under concurrent I/O, which is what databases and build systems generate. For a static site, you won't notice. The page explains when it matters and how to measure it with fio.
Managed vs unmanaged
Managed hosting means someone else patches, monitors and fixes the operating system and often the application stack. Unmanaged, or self-managed, means you do. FastNode is self-managed. The page lists exactly what our support covers, what it doesn't, and the questions to ask yourself before choosing a server you'll run on your own.
How to use these pages
Start with the constraint that hurts most today. If your bill is the problem, read the serverless and dedicated pages, which are mostly about cost curves. If performance is the problem, read the storage page and then measure: iostat, vmstat and mtr answer most "is it the server?" questions in ten minutes. If the problem is time, and nobody has enough of it to look after an operating system, read managed versus unmanaged first, because no amount of NVMe fixes an unpatched box.
Each comparison stands alone, uses plain technical terms and avoids naming other providers. Where numbers appear, they describe the technology in general or our published plan specifications, not benchmarks we've run and dressed up. Where we can't back a claim, we leave it out.
What we won't tell you
We won't claim our servers beat a named competitor, because a benchmark taken on one day on one host tells you very little about the next. We won't invent uptime figures or customer counts. And we won't pretend a single London region is the right home for an audience spread across continents. If a page here talks you out of ordering, it did its job.
A quick decision table
| If you need… | Lean towards |
|---|---|
| Pay only when code runs, near-zero idle cost | Serverless |
| Long-running processes, websockets, local disk | VPS |
| All cores flat out for months, 256 GB+ RAM | Dedicated |
| Your own kernel, modules or a non-Linux OS | Virtual machine (VPS) |
| Packing many small services onto one machine | Containers on a VPS |
| Concurrent random I/O: databases, CI caches | NVMe-backed storage |
| No time or skill to run Linux yourself | Managed hosting |
Ready to check sizes? The pricing page lists all six London plans. Not sure what to run on them? Browse the use cases.