The managed vs unmanaged server decision is usually framed as a budget question — pay more for management, or save money and do it yourself. That framing is wrong in a specific way: unmanaged hosting is not cheaper, it is differently priced. The cost moves from an invoice to a calendar, and calendars are harder to read.
Here is what each side actually covers, the labour arithmetic, and the cases where each is the right answer.
What "unmanaged" means in practice
Unmanaged means the provider owns the hardware, the network and the power. Everything above the operating system is yours from the moment the server is provisioned.
| Responsibility | Unmanaged | Managed |
|---|---|---|
| Hardware failure, network, power | Provider | Provider |
| OS installation | Provider (image) | Provider |
| Kernel and package patching | You | Provider |
| Firewall rules and hardening | You | Provider |
| Monitoring and alerting | You | Provider |
| Backup configuration and testing | You | Provider |
| Performance tuning | You | Provider |
| 3am incident response | You | Provider |
| Application code and config | You | You (mostly) |
The last row is the one people misread. Managed hosting does not mean somebody else runs your application. It means somebody else keeps the platform underneath it healthy. Your deployment is still your deployment.
The labour arithmetic
The honest way to compare is to price the hours. A single production Linux server, run properly, needs:
- Patching: roughly monthly, plus out-of-band for anything critical. Half an hour to two hours each time, including verifying nothing broke.
- Monitoring upkeep: alerts drift, thresholds need tuning, checks need adding as the service grows. An hour or two a month once the initial setup is done.
- Backup verification: a restore test worth the name is an hour a quarter. Skipping it is the most commonly skipped task in infrastructure and the most expensive one to skip.
- Security review: open ports, users, keys, certificate expiry. An hour a quarter, minimum.
- Incidents: unpredictable by definition, and always at the worst time.
Call it four to six hours a month in the steady state for one server, excluding incidents. Price that at whatever an engineer's hour is worth in your organisation and compare it against a fixed monthly management fee. For most teams the fee wins on cost alone — and that comparison ignores the part that actually matters.
The part the hours calculation misses
Patching does not happen. Not because engineers are lazy, but because patching a production server is unglamorous work that competes with shipping features, and it loses that competition every sprint until an audit or an incident. The single biggest practical difference between managed and unmanaged fleets is that managed servers are patched on a schedule and unmanaged ones are patched when somebody remembers.
On-call is a person, not a process. If one engineer owns the server, that engineer cannot take an uninterrupted holiday. That is a real cost that never appears in a comparison.
Expertise is thin at the edges. Most developers can run a server well enough. Fewer can tune MySQL's buffer pool against a real working set, size PHP-FPM against actual memory, or read a kernel trace at 3am. Those hours are rare and expensive when you need them.
Compliance turns discipline into a requirement. Under NIS2, patching cadence, monitoring and incident handling stop being good practice and become things an organisation must be able to demonstrate. "We patch when we get to it" is not a defensible answer — see what NIS2 asks of hosting providers for the detail.
NexonHost's managed infrastructure services cover exactly this band: monitoring of CPU, RAM, disk and network; OS and software updates applied on schedule; firewall configuration, intrusion prevention and vulnerability patching; performance tuning; automated backups with recovery assistance; and proactive incident response — at a fixed monthly rate, on Linux, Windows or custom setups. For multi-site or hybrid estates, IT infrastructure management extends the same discipline across locations and clouds under one framework.
When unmanaged is genuinely the right call
Being fair to the other side, because plenty of good operations run unmanaged.
You already have an ops function. If there is a team with on-call rotation, configuration management and a patching cadence that actually runs, buying management is buying something you have. Unmanaged is correct.
The stack is unusual. A custom kernel, an uncommon distribution, a bespoke storage layer, a workload that violates every general best practice for good reasons. Management contracts are built around standard stacks; a genuinely unusual one gets less value from them.
Infrastructure is the product. If you sell hosting, run a platform, or your engineering differentiation is how the servers are run, outsourcing that is outsourcing the thing you sell. Agencies reselling to clients are the interesting middle case here — see reseller hosting margins for when a pool beats a machine.
It is genuinely non-critical. A build agent, a scratch box, an internal tool three people use. Not everything needs a maintenance contract, and pretending otherwise wastes money.
You want to learn. A legitimate reason, on a machine that is allowed to break.
The middle option people forget
Managed and unmanaged are not the only two positions, and the useful third one is under-sold: keep root, buy the maintenance. You retain full access and deploy how you like; the provider handles patching, monitoring, hardening and backups underneath.
This suits the common case of a team that is perfectly capable of running a server and has more valuable things to do — the team whose objection to managed hosting is loss of control rather than cost. If that is the objection, ask specifically whether root access is retained, because with most providers including NexonHost it is.
Questions to ask before signing either way
For a management contract:
- What is the patching cadence, and who decides what is urgent?
- What is monitored, and who receives the alert at 3am? A dashboard nobody watches is not monitoring.
- Are backups configured, and are restores tested? Ask when the last test restore was.
- Is root access retained?
- What is explicitly out of scope? Application code and third-party software usually are, and that boundary should be written down before an incident rather than during one.
- What is the response time, and is it a target or a commitment?
For unmanaged, ask yourself the same list. If you cannot answer all six about your own servers today, that is the actual result of the comparison.
Choosing, briefly
Run unmanaged when you have the capability and the workload deserves your attention. Buy management when the server is production, the team is small, and the alternative is patching that happens irregularly and a restore nobody has tested.
The failure mode of unmanaged hosting is not a bill. It is a machine that has been running fine for eighteen months, unpatched, with a backup job that has been failing silently since March. That server is cheap right up to the day it is the most expensive thing you own.
Related reading: VPS vs dedicated server, the dedicated server DDoS protection checklist, and offsite backup storage sizing.
Sources
- Service scope, coverage and support arrangements as published on nexonhost.com on 7 September 2026; check the product page for current terms.
- Labour estimates in the "labour arithmetic" section are illustrative planning figures, not measured benchmarks — substitute your own observed effort before using them in a business case.
