A few years ago a systems administrator could spend a working life inside one server room. Now the same person may hold a directory that syncs to Entra ID, a virtualisation platform whose licensing has just changed, cloud virtual machines and a Microsoft 365 tenant, with ransomware as the disaster they plan for most. The system administrator skills that count today span all of those, and they lean far harder on identity and recovery than on racking hardware.
Few candidates will be deep in everything, so the aim is to separate real depth from a list of product names. This post sets out nine core skills, each with a test you can run, the habits that make an administrator trustworthy with privileged access, what is gaining and losing importance, a level matrix and an assessment plan that goes beyond the usual stories about outages.
Core technical skills for a systems administrator
Active Directory internals
Anyone can create a user. A strong administrator understands the FSMO roles, sites and services, and why Kerberos depends on accurate time. They also know how to retire old protocols, as Microsoft continues moving customers away from NTLM. How to check it: ask what symptoms appear when a domain controller’s clock drifts well out of step, and which role holder should anchor time in the domain. Candidates with real experience link failed Kerberos tickets to the PDC emulator without much prompting.
Group Policy diagnosis
Policies fail quietly, which makes this a good measure of method. Skilled people understand link order, inheritance, security filtering and WMI filters, and they know which results come from the client’s view. How to check it: say a policy appears correctly linked but one group of computers ignores it. Ask for the steps. Good answers start with gpresult on an affected machine, then check filtering and whether the setting belongs to the user or the computer side.
File services and permissions
Share and NTFS permissions cause a steady flow of escalations, and a messy structure becomes an audit finding. Competent administrators assign access through groups, not individuals, and can untangle inherited rights. How to check it: ask why a user who was added to a group an hour ago still cannot open a share. The usual reason is that group membership sits in the logon token, so the person needs a new session, and an experienced administrator says so at once.
Linux service management
Red Hat, Rocky and Ubuntu estates call for comfort with systemd, journalctl, LVM and package repositories, plus SELinux on the Red Hat family. Turning SELinux off is the giveaway of someone who never learned it. How to check it: describe a web service that fails after its content moves to a new directory, although file permissions look fine. Ask what they would inspect. Look for the audit log, file contexts and restorecon.
Virtualisation
VMware vSphere remains widespread, though Broadcom’s licensing changes have led many companies to price Hyper-V and Nutanix seriously. Administrators need to understand clusters, storage, snapshots and live migration. How to check it: ask what can go wrong when a virtual machine snapshot is left running for months. Datastore growth, slower performance and a painful consolidation are what you want to hear, along with the point that a snapshot is not a backup.
Backup design against ransomware
Veeam and Commvault jobs are the visible part. The design around them decides whether anything survives. Strong administrators separate backup credentials from the production domain and keep an immutable or offline copy. How to check it: say that an attacker holds domain admin rights and has encrypted every joined server. Ask what would survive in their design, then ask which services they would bring back first. Identity and name resolution should come early in that order.
Microsoft 365 and identity
Administrators often inherit the tenant along with the servers, which brings Conditional Access, MFA, licensing and hybrid sync through Microsoft Entra Connect Sync or, increasingly, Cloud Sync. A single careless policy can lock everyone out. How to check it: ask how they would introduce a Conditional Access policy that requires compliant devices. Good answers include report-only mode, a pilot group and emergency accounts excluded from the policy.
Privileged access and hardening
This role holds the keys, so its habits shape the security of the estate. Good administrators use separate admin accounts, restrict where those accounts can log in and rotate local administrator passwords with Windows LAPS. How to check it: ask why logging in to an ordinary workstation with a domain admin account is dangerous. A sound answer covers credentials exposed in memory and how an attacker on that workstation can reuse them to move across the network.
Safe scripting
PowerShell and Bash save time, and they also make it easy to break many servers at once. Skill shows in error handling, logging and a dry run before anything destructive. Ansible helps once an estate outgrows individual scripts. How to check it: hand over a short script that deletes old files from a path held in a variable. Ask what happens if that variable comes back empty. Good candidates add validation and test with WhatIf or an echo first.
Working skills that separate good from average
- Discretion. Administrators can read mailboxes, salary folders and board papers. The good ones are careful with that power, log what they do and raise it when access is broader than a task requires.
- Writing things down as they go. A small IT team can lose years of knowledge when one person resigns. Strong administrators leave a trail: why a server exists, what depends on it and how to rebuild it.
- Handling people during an outage. Staff want to know when things will work again. Good administrators send plain updates at sensible intervals and keep managers away from the person fixing the fault.
- Retiring things. Estates fill up with servers nobody dares to switch off. Strong administrators trace the owner, plan a scream test with care and actually complete the decommissioning.
What is rising and fading for administrators in 2026
Mattering more. Identity has become the main security boundary, so Entra ID, Conditional Access and hybrid sync now carry the weight that perimeter firewalls once did. Recovery from a deliberate attack differs sharply from recovery after a failed disk, and administrators need to plan for it. Virtualisation choices have reopened, which rewards people who can compare platforms honestly. AI assistants write serviceable PowerShell quickly, so reading a script and spotting what it would do to production has become a core skill. Cloud cost hygiene and Azure Arc or similar hybrid management also appear more often in everyday work.
Mattering less. On-premises Exchange administration is shrinking as mail moves to Exchange Online. Microsoft has deprecated WSUS, so patching is moving towards Intune and Azure Update Manager, although many estates will run WSUS for some time yet. Building physical servers by hand, and memorising obscure registry paths, count for less. Long, hand-followed build documents are giving way to templates and configuration management.
Systems administrator skills matrix by level
| Skill area | Junior | Mid-level | Senior |
|---|---|---|---|
| Active Directory | Accounts, groups, resets | Replication, sites and policy diagnosis | Forest design, tiering, protocol retirement |
| Linux | Services, logs and packages | LVM, SELinux and troubleshooting | Hardening baselines and fleet standards |
| Virtualisation | Manages individual machines | Clusters, storage and capacity | Platform choice and migration plans |
| Backup and recovery | Checks jobs, runs simple restores | Owns restore tests and retention | Designs for ransomware and sets recovery order |
| Microsoft 365 | Licences and mailbox tasks | Conditional Access and sync | Tenant governance and identity strategy |
| Security | Uses separate admin account | Applies LAPS and hardening baselines | Designs privileged access model |
| Automation | Runs and adapts scripts | Writes safe, logged scripts | Moves the estate to configuration management |
How to assess systems administrator skills in a hiring process
Everyone can tell a polished outage story, so build in tasks that cannot be rehearsed. If the role itself is still unclear, begin with the systems administrator job description, which includes the broader screening questions. Then try the following.
- Give them a broken virtual machine. A small lab with a stopped service, a wrong DNS entry and a full volume shows method quickly. Ask them to talk aloud as they work.
- Audit a small environment on paper. Share an anonymised export listing admin group members, password settings and backup jobs. Ask for the three things they would fix first and why. Priorities reveal more than product knowledge.
- Write the handover note. After the lab, ask for a short note a colleague could act on. In small teams this matters as much as the fix.
- Discuss a refusal. Ask about a time they pushed back on a request for access or an unsafe change. You want someone who can say no without creating a feud.
Red flags on a CV: a long product list and no estate size; “responsible for backups” with no mention of restores or testing; Linux listed alongside only desktop work; every duty written in the passive voice; and years spent in one environment with no sign of anything improved.
Frequently asked questions
Must a systems administrator be equally strong on Windows and Linux?
Rarely. Most people lean one way, which is fine if the lean matches your estate. Expect depth on your main platform and working competence on the other, enough to read logs, manage services and know when to ask for help.
How much networking should an administrator know?
Enough to rule the network in or out: subnets, VLANs, DNS, routing basics and firewall logic. They should be able to prove that a port is blocked and hand a clear finding to a network engineer, who owns design and changes.
Will AI tools take over routine administration?
They already help with scripts, log summaries and documentation. Accountability stays with a person, though, since someone has to judge whether a suggested command is safe on a live server. Expect fewer repetitive tasks and more review.
How good should PowerShell skills be?
For a Windows-heavy mid-level post, a candidate should write a script with parameters, error handling and logging, and query remote machines. They need not build modules. Being able to read and correct someone else’s script is the better test.
We screen on real outages, restores and upgrades, plus the scripting that takes routine work off your team, and Zilo AI can help you hire systems administrators in India on contract, contract-to-hire or permanent terms. Our IT hiring hub shows the related Linux, Microsoft 365 and backup specialisms. Describe your estate to us and tell us where the gap sits.