Why pre-screen storage engineers before the technical interview
Storage is invisible until it is catastrophic, which makes the discipline hard to assess by reputation. An engineer can run a healthy environment for years because nothing stressed it, and never discover whether their backups restore or whether their capacity plan survives a growth spike. The two questions that separate candidates are whether they have performed a real restore under pressure and whether they can trace latency across the boundary between storage, network and host. Neither appears on a resume.
What actually matters when screening Storage Engineer candidates
- 01
Technical depth
Check depth across block, file and object: NetApp ONTAP or Pure arrays, Fibre Channel zoning on Brocade fabrics, multipathing, snapshots, SnapMirror or async replication, S3 tiering.
- 02
Work that shipped
Probe migrations and builds they owned: array refreshes, datastore evacuations, vSAN or Ceph clusters, capacity added in TB, downtime windows honoured, DR failover tests actually executed.
- 03
Diagnosis under uncertainty
Test how they chase latency spikes, path flapping, failed disk rebuild storms or backup windows overrunning, using array counters, esxtop, iostat and fabric error logs.
- 04
Working across the org
Assess work with DBAs, virtualisation and backup teams on capacity forecasts, tiering policies, RPO and RTO commitments, change advisory boards and vendor support escalations.
Pre-screening questions to ask Storage Engineer candidates
12 questions grouped by what they test. Ask the same set in every screen and score answers on a consistent scale, or send them as an async video screen and compare answers side by side.
Storage depth
3 questions01What is your experience with Storage Area Networks and network-attached storage?
Listen forBoth protocols and their trade-offs described from operating them, with the fabric or network side included rather than the array alone.
Names both without distinguishing their use cases, or experience limited to provisioning volumes in a console.
02Do you have experience with flash storage or NVMe?
Listen forPractical understanding of where the bottleneck moved once media stopped being the limit, and a workload where flash did not solve the problem.
Treats flash as a universal fix, or no awareness that the bottleneck moves to the network or the host.
03What is your knowledge of distributed storage systems such as Ceph or similar platforms?
Listen forOperating experience with rebuild behaviour and failure domains understood, rather than familiarity with the architecture in principle.
Describes distributed storage conceptually, or has never seen a cluster rebuild after losing a node.
Designing for growth
3 questions04Can you discuss a project where you designed and implemented a storage strategy from scratch?
Listen forRequirements gathered before technology chosen, with growth projections, tiering decisions and something deliberately left out of scope.
A design built around a preferred vendor, or capacity planned with no growth model behind it.
05What strategies do you use to manage large volumes of data?
Listen forTiering and lifecycle policies applied in practice, with data actually moved or deleted rather than everything kept on primary storage.
Growth handled by buying more capacity, with no lifecycle policy ever enforced.
06Can you discuss your experience with compression, deduplication and thin provisioning?
Listen forReal ratios achieved on real workloads, plus awareness of the overcommitment risk thin provisioning creates and how they monitored it.
Quotes vendor ratios rather than observed ones, or thin provisioning used with no monitoring of actual consumption.
Latency diagnosis
3 questions07How do you monitor and optimise storage system performance?
Listen forA latency investigation that crossed into host or network, with the queue depths or path issues they examined before blaming the array.
Examines array metrics only, or concludes storage is healthy while users continue reporting slowness.
08What is your experience with RAID and other aspects of disk management?
Listen forRAID levels chosen for a workload with rebuild time and risk during rebuild considered, not just capacity and redundancy.
Applies one RAID level everywhere, or no awareness of the exposure window during a rebuild.
09What experience do you have with storage-related scripting or automation?
Listen forScripts they wrote for reporting or provisioning, with a repetitive task they removed and roughly what it saved them.
All work performed through the vendor console, or automation claimed with no example of a script they wrote.
Restores they have run
3 questions10Do you have experience managing backups, replication and disaster recovery?
Listen forA restore they actually performed with the recovery time stated, plus what did not work as documented and what changed afterwards.
Reports backup jobs completing successfully with no restore ever tested, or recovery objectives never validated.
11Can you explain how you have managed data protection and recovery?
Listen forRecovery point and recovery time objectives agreed with the business, with a case where the achievable figure was worse than the promised one.
Objectives inherited with no validation, or protection described without any conversation about what the business needs.
12Have you been part of a data migration project? What role did you play?
Listen forA migration with the cutover approach and rollback path described, plus what was reconciled afterwards to confirm nothing was lost.
Migration described with no verification step, or a cutover with no way back if it failed.
How to score responses
Score every candidate on the same four criteria immediately after the screen. At this stage you are shortlisting for panel interviews, not making the final call.
Technical depth
35%5Names array models, LUN and aggregate layouts, zoning schemes and replication modes, with clear reasoning on IOPS, latency and RAID protection levels.
Work that shipped
30%5Describes a completed migration with petabytes moved, cutover plan, rollback path, and measured latency or utilisation outcomes afterwards.
Diagnosis under uncertainty
20%5Walks a real incident from symptom to root cause across host, fabric and array layers, citing the counters and logs that isolated it.
Working across the org
15%5Shows negotiated capacity and RPO agreements, clear runbooks handed to operations, and vendor cases driven to resolution without stalling projects.
Storage is invisible until it is catastrophic, so a quiet environment proves very little. A one-way video screen lets you hear whether a candidate has ever restored data for real.
Try it on HirevireScreening FAQ
Process basics
How long should a pre-screening round for a storage engineer take?
Fifteen minutes across eight to ten questions, answered async. Enough to confirm the technologies they have actually operated, hear a performance investigation, and find out whether they have ever restored data for real.
How much should vendor certifications count?
They confirm training on a platform and often matter for support contracts. They tell you nothing about whether someone has recovered from a failure. Ask about restores and migrations directly rather than treating the certification as a proxy.
Evaluating answers
What is the strongest signal when screening a storage engineer?
A restore they actually performed. Engineers who have done it describe the recovery time, what did not work as documented, and what they changed afterwards. Candidates who describe backup jobs completing successfully have not tested anything.
How do I judge performance troubleshooting answers?
Listen for whether they looked beyond the array. Latency complaints usually turn out to involve the host, the network or a noisy neighbour. An engineer who only ever examines storage metrics will keep concluding that storage is fine while users disagree.
Go deeper on this role
Sanat has been hiring since 2012 and watching the recruitment industry change up close ever since, and turned that screening process into Hirevire's video screening platform. LinkedIn
