◇ Independent. In-depth. Infrastructure first.

Methodology · Resilience

Compare shared, managed and isolated hosting routes

Evaluate responsibility, limits, control, support and recovery rather than choosing from a feature grid.

Shared managed and isolated hosting responsibility comparison

Shared hosting, managed hosting and a VPS do not describe the same dimension of a service. “Shared” concerns how infrastructure is allocated; “managed” describes work the provider agrees to perform; a VPS provides a virtual server environment. A managed service can still share infrastructure. Start with the responsibility you can realistically take on, then compare capacity and price. This guide is a documentary analysis, not a speed test or a ranking of providers.

Identify who operates each layer

Write down five layers: physical infrastructure, operating system, web server and database, application, and content. Ask who installs updates, watches for incidents and restores data at each layer. A statement such as “security included” does not answer whether an outdated WordPress plugin will be fixed or whether a failed checkout will be diagnosed.

Hostinger’s documentation describes its Web and Cloud hosting as managed at the server level and its VPS offering as self-managed. It also distinguishes those server tasks from the customer’s website responsibilities. This is one documented provider example, not a definition that automatically applies to every host or future product.

Choose a route for the workload, not the label

An editorial site run by one non-technical person

A conventional managed web plan may be a practical starting point when the application is supported and traffic is modest. Check the renewal price, backup scope, restoration procedure and support boundaries before purchase. Root access is not an advantage if nobody will use it safely. A useful constraint is that another trusted operator should be able to restore service without reconstructing an undocumented server.

A business site with regular publishing and enquiries

Prioritise staging, reliable delivery of enquiry messages, safe updates and recoverable data. Ask whether backups include both files and the database and how recent a restored copy can be. Separate website hosting from email hosting in the checklist: moving the site must not accidentally remove mailboxes or change their DNS records.

An application needing custom services

A VPS can be appropriate when the application requires system-level configuration that a standard plan does not allow. Someone must then own patching, firewall rules, credentials, database recovery and monitoring. A control panel or installation template can simplify setup without taking over all those responsibilities. If that work has no owner, reduce the technical scope or compare a managed alternative.

Compare limits that affect the actual site

Do not choose from disk space alone. Record application versions, memory and execution limits where disclosed, database constraints, background job support and the conditions for resource throttling. Ask what happens when a limit is reached: does the site slow down, reject requests, incur charges or require an upgrade? An unclear answer is a reason for clarification, not permission to assume unlimited use.

Traffic totals also hide differences. Ten thousand cached article visits are not equivalent to ten thousand uncached account or checkout requests. Write down the busiest useful journey and its dependencies. Treat any proposed capacity as a hypothesis until representative testing and real observations support it.

Calculate cost across the full commitment

Use the initial amount charged, renewal amount, commitment length, taxes where applicable and necessary extras. Add the cost of someone maintaining the service. A low introductory monthly equivalent can accompany a much larger upfront payment. Record cancellation and refund conditions separately instead of assuming that all trial or refund arrangements are interchangeable.

For illustration only, suppose option A costs 120 currency units per year and requires two hours of monthly maintenance. At an assumed internal cost of 25 units per hour, that is 720 units annually before other costs. Option B costs 300 annually and needs half an hour a month: 450 units under the same assumption. These are invented arithmetic inputs, not provider prices or a forecast. Change them to your own estimates.

Verify recovery, not just the existence of backups

Define the largest acceptable data loss and interruption for the site. An information page and an order database may need different recovery arrangements. In a test environment, restore the files and database together, verify login and important journeys, and identify the person who can repeat the procedure. Keep any off-site copy protected and check how it is removed when no longer needed.

Do not mistake a successful control-panel message for a complete recovery exercise. Check uploaded images, scheduled tasks, email delivery and DNS-dependent integrations. Record what has actually been tested and what remains assumed. Never perform a destructive restoration rehearsal on the only production copy.

Questions to ask before committing

  • Which server and application tasks are included, and which remain mine?
  • What does the plan cost at renewal and when upgrading?
  • Can I restore a sample backup without additional assistance or charges?
  • What happens during excessive resource use or an outage?
  • How do I export the website, database and email if I leave?

The best choice is the smallest service that meets the workload and has an owner for every critical task. More control is useful only when it solves a real requirement and the team can maintain it.

Documentation checked: 26 September 2026. Provider reference: Hostinger Help Center, “What is managed hosting in Hostinger”. No live prices, affiliate offers, uptime claims or hands-on benchmark results are presented.

Next practical steps