In this guide: What is server isolation and why it matters on shared hosting: how shared hosting works, the bad neighbor risk, isolation technologies, questions to ask…
Shared hosting is affordable because many websites share one server. But that raises an important question: if another website on your server gets hacked or overloaded, can it affect yours? The answer depends on server isolation.
On a shared server:
Without proper isolation, accounts can interfere with each other — in performance and in security.
Performance risks:
Security risks:
wp-config.php with database passwords) if permissions and isolation are weak.Reputation risks:
Good hosts use several layers to separate accounts:
Separate system users and file permissions
Each account runs under its own user, and PHP runs as that user (via PHP-FPM or LSAPI), which can support separation when permissions, runtime configuration and the surrounding controls are implemented correctly. It is not an absolute security guarantee.
CloudLinux with LVE and CageFS
Widely used on shared hosting:
Containers
Some hosts run each account or site in its own container, with separation that depends on namespaces, privileges, mounts, resource controls and the surrounding configuration. Containers are not automatically equivalent to independent virtual machines.
Per-account PHP processes and open_basedir restrictions
Limit which directories PHP scripts can access.
Mandatory security modules
Server-level WAFs and malware scanners (e.g., Imunify360) can provide detection and traffic filtering, but do not replace account separation or guarantee containment.
Separate IPs (optional)
Some plans offer dedicated IPs, useful for specific email or legacy needs.
Confident, specific answers are a good sign. Also check reviews for recurring complaints about slowdowns or infections. See Is Your Hosting Secure? 12 Features to Check.
Consider moving to a VPS, cloud server, or managed hosting with stronger isolation if:
A VPS can provide a different administrative and virtualization boundary; actual resource guarantees depend on the plan and architecture — see How to Move From Shared Hosting to a VPS. Managed WordPress hosting often uses container-based isolation per site.
For many small sites, appropriately configured shared hosting may fit the risk and workload — as long as you also keep your own site secure.
Further reading: WordPress Advanced Administration Handbook
Start with a concrete boundary rather than a product label. Are you asking whether one account can read another account’s files, whether a busy workload can consume shared capacity, or whether an administrator can access customer data? These are different questions and require different evidence.
Write the concern in practical terms. A small agency may want separate credentials and recovery for each client. A store may need predictable operation during busy periods. A publisher may need to limit the effects of a compromised extension. No single marketing phrase establishes that all of those requirements are met.
Keep the assessment proportional to the actual site. Do not assume that every small publication needs a dedicated machine, or that every sensitive workload is suitable for shared hosting. Define the data, external obligations and operational needs before comparing infrastructure categories.
A resource limit and a file-access boundary solve different problems. A CPU limit may constrain a workload without establishing which files that workload can read. Conversely, appropriate file permissions do not ensure that storage or network capacity cannot become a bottleneck.
Ask the provider to explain the relevant controls separately. Record what is limited, the scope of the limit and what happens when it is reached. A plan with a resource allowance needs more explanation than a phrase such as unlimited websites, especially when several applications share the same account allocation.
Our shared hosting to VPS guide helps frame a migration decision. Moving categories without identifying the original constraint can add administrative work without resolving the actual performance problem.
CloudLinux’s introduction distinguishes resource management through LVE from the per-user filesystem role of CageFS. Those descriptions explain the technology’s purpose; they do not prove that a particular hosting plan enables or configures every available component.
Request plan-specific confirmation rather than treating a software logo as evidence. Ask whether the controls apply per account, per site or across a reseller allocation. Keep the provider’s answer and date with the comparison so a later reviewer can see what was actually confirmed.
The CloudLinux limits documentation is useful for understanding the categories involved. Do not copy example configuration values into a hosting comparison as if they were the limits of the plan you are considering. Example values and purchased entitlements are different evidence.
Containers can support useful separation, but their configuration and shared dependencies matter. Docker’s container explanation notes that containers on a machine share its kernel. That is an important distinction when comparing them with a separate virtual-machine boundary.
Docker’s security guidance discusses risks involving privileges, mounts and incomplete isolation. Do not claim that a container automatically prevents every cross-workload incident. Equally, the existence of shared dependencies does not by itself diagnose a particular provider as insecure.
Ask what the boundary is intended to contain and who maintains the underlying environment. A responsible comparison distinguishes documented architecture, provider statements and independent tests. Do not describe a hosting service as tested simply because its marketing page uses a familiar technology name.
The PHP manual explicitly describes open_basedir as an additional safety net rather than comprehensive security. Avoid presenting a directory restriction as a complete substitute for operating-system permissions and the surrounding hosting controls.
For a managed hosting customer, the useful question is how the provider implements and maintains separation, not which settings you can disable. Do not weaken controls or probe another customer’s account to test a claim. Review documentation and use an authorized assessment process when stronger evidence is required.
Keep the customer’s own application responsibilities visible. Vulnerable extensions, exposed credentials and excessive administrator access can create problems inside the account even when the provider’s cross-account boundary works as intended. Infrastructure and application maintenance remain complementary responsibilities.
Several sites under one hosting account may share credentials, permissions and recovery scope. Separate domain names do not automatically create separate administrative boundaries. Ask what happens if one application or account credential is compromised and which other resources the same identity can access.
An agency should consider client ownership, support access and departure as well as technical separation. A client leaving the service needs a controlled export and handover. A shared administrator password and a single unlabelled backup can make that process difficult even without an attack.
Our hosting for agencies guide and reseller planning guide provide complementary questions. Do not assume that a reseller plan automatically establishes the same boundary as a separately managed virtual server.
A slow page is not proof of a noisy neighbour. The application, database, cache, network and client device can all affect the observed journey. Begin with the affected page, time and conditions, then compare available resource and application evidence.
Ask whether the account reached a documented limit and what the provider observed. Keep an unknown category when evidence is missing. Avoid assigning fault to another customer or a shared server based only on an intermittent delay seen from one device.
Review our Core Web Vitals explanation and caching guide for related investigation. A benchmark needs its method and limitations; a hypothetical example should never be presented as a measured hosting result.
Isolation does not eliminate the need for backups and recovery ownership. A problem contained within one account can still destroy that account’s files or content. Check what the backup covers, where it is held and who can restore it.
Recovery scope matters for multiple sites and live transactions. A broad restore can affect newer orders, users or other applications in the same database environment. Ask whether the relevant data can be recovered safely rather than assuming that an account-level restore fits every incident.
Our backup guide and WordPress update workflow explain complementary acceptance questions. Keep a dated recovery record and a suitable rehearsal, without claiming that a backup status alone proves a successful restoration.
A different hosting architecture can change the boundary and control available, but it also changes responsibilities. Determine who handles operating-system updates, application maintenance, monitoring and incident response. Unmanaged access requires skills and time that should be included in the decision.
Read resource terms carefully. Dedicated CPU, shared CPU, memory allowances and burst conditions describe different arrangements. Do not claim guaranteed performance merely because a product is called VPS or cloud. Ask for the plan’s actual terms and assess them against the workload.
Managed also needs a written definition. Our managed WordPress guide distinguishes service responsibilities from assumptions. A provider may maintain one layer while leaving custom code, content and integration support to the site owner.
Record each requirement, the source of the answer, its date and any unresolved question. Separate provider claims from an independent assessment. A confident support response can be useful, but it is not the same as a verified audit of the environment.
Choose an action that matches the evidence: retain the current plan, adjust an application issue, request clarification or plan a migration. Avoid treating an upgrade as the automatic answer to every slowdown or security concern. Define the acceptance checks before committing to a move.
Review the record when the workload, data or operating team changes. The appropriate arrangement for a small publication may differ from the next stage of the business. An evidence-based decision is more useful than labelling an entire hosting category perfectly safe or inherently insecure.
Promotional recommendation: KAWEDMA offers hosting services that you can consider when comparing a suitable plan. Request current, plan-specific answers about account separation, resource limits, backups and management scope. This recommendation does not assert that a particular isolation technology or resource guarantee is included without confirmation.
Affiliate disclosure: Hostreta may earn a commission if you purchase through these links. Compare hosting.com’s managed VPS offering and its unmanaged VPS offering only where that category matches your requirements. Verify current resources, management boundaries and renewal terms. Neither link is evidence of a measured security or performance result.
Consider a fictional agency with three client websites. The team wants to understand whether one site’s incident could affect the others and whether the current plan suits the workload. This example is a decision exercise, not an assessment of a real host or evidence that an account has been compromised.
The first row names the administrative boundary. Are the sites controlled by one account, separate accounts or a different arrangement? The team records who can access each site and which credentials are shared. The answer needs the provider’s actual configuration and the agency’s own access practices, not a conclusion drawn from the number of domain names.
The second row concerns resource allocation. The team asks which limits apply to each site or account, whether a reseller allocation also constrains them and how an exceeded limit is reported. An example value from a software manual is not copied into the plan column. Any missing answer remains marked as unconfirmed.
The third row concerns recovery. The agency asks whether it can restore one client’s files and database without overwriting another client’s current work. It records backup ownership and the supported recovery process. A broad account restore may not fit the desired separation, even when the underlying hosting access boundary works as intended.
The fourth row concerns application maintenance. A compromised administrator account or vulnerable extension can affect the resources that identity already controls. The team records update ownership, authentication and response responsibilities. It does not treat infrastructure separation as permission to neglect the applications running inside the boundary.
Finally, the agency compares the proposed alternatives against the same requirements. A more expensive plan is not awarded a higher security rating merely because of its price. The team separates documented capabilities, provider statements and unanswered questions, then chooses a next action appropriate to the evidence.
After any change, verify the operational result with an authorized process. Do not attempt to read another customer’s files or weaken restrictions to test the marketing claim. If stronger assurance is necessary, arrange an appropriate assessment with a defined scope and responsible parties. The worksheet helps organize that conversation without pretending to be an audit or compliance certificate.
Keep the completed record with the service agreement and review it when a new client, integration or data requirement changes the workload. A clear explanation of the isolation boundary and its limitations is more valuable than an unsupported promise that a category of hosting prevents every possible incident.
Recheck documented isolation requirements when ownership, workload or support responsibilities change.
We may earn a commission on purchases made through these links, at no extra cost to you.
Don’t miss the chance to participate in the biggest giveaway of this year.