Server Hardware and Storage Arrays
Specify and operate server platforms with redundant power, RAID controllers, out-of-band management, and shared storage.
- Certification
- CompTIA Server+
- Recommended study time
- 5h 35m
- Status
- Not started
Recommended study time
About 5h 35m in total, measured from the material on this page. At your session length of 45 minutes that is 8 sittings.
- Read the lesson23 min
About 3,034 words at a careful technical reading pace.
- Second pass with notes14 min
Re-read the harder parts and write your own notes.
- Recall from memory12 min
2 written recall questions.
- Practice decision12 min
One applied decision with feedback.
- Teach it back20 min
Write the topic in your own words.
- Real-world scenario15 min
Read the situation and justify your decision in writing.
- Hands-on practice2h 40m
Labs, commands and configuration until you can do it unaided.
- Spaced review1h 20m
4 short review sessions spread over the following weeks.
Learning objectives
- Compare rack, blade, and converged platforms and their operational trade-offs.
- Explain redundancy features and where single points of failure remain.
- Use out-of-band management for remote diagnosis and recovery.
Start here
About 8 minutes of reading, in 10 short parts.
Server hardware is built around a single goal that desktop hardware mostly ignores: staying running when a part fails. Understanding redundancy, RAID, and remote management is what lets you keep a service available through routine component failures instead of treating every failed drive as an emergency outage.
Where you meet it: A drive fails in a production array, and whether that is a non-event or a disaster depends entirely on decisions made when the array was built.
The lesson, part by part
Open one part at a time. Each part stands on its own, so you can stop and come back.
A desktop computer is built to be affordable, and if something breaks, you fix it or replace the whole machine over a weekend. A server is built on the opposite assumption: something will break while it is running, and it must keep working anyway, because a database or website going down at 2 a.m. is not acceptable. This means servers duplicate the parts most likely to fail — power supplies, fans, and disks — so that losing one does not stop the whole machine.
Storage arrays extend this idea to disks specifically. Instead of storing your data on a single disk that could fail and take everything with it, RAID spreads or duplicates data across several disks using different strategies, trading off cost, speed, and how many disks can fail before data is actually lost. Out-of-band management is like having a second, independent phone line into the building that still works even if the main phone system is down, letting an administrator reach and control the server even when its main operating system has crashed.
Key ideas
If you remember nothing else from this topic, remember these.
- RAID levels trade capacity, performance, and fault tolerance against each other, and choosing the wrong one for a workload creates either wasted cost or unacceptable risk.
- RAID 1 mirrors data across drives for redundancy with no performance write penalty benefit beyond redundancy, while RAID 5 and 6 use parity to tolerate one or two drive failures respectively at lower capacity cost than mirroring.
- RAID is not a backup; it protects against a drive failure, not against accidental deletion, corruption, or a site-level disaster.
- Hot spares let an array begin rebuilding automatically the moment a drive fails, reducing the window during which the array runs in a degraded, more vulnerable state.
- Redundant power supplies and NICs address a different failure category than RAID, protecting against component failure rather than data loss.
- mdadm manages Linux software RAID arrays and reports their state, including degraded arrays that still function but have lost their fault tolerance.
Responding to a degraded RAID 5 array on a file server
A worked example, step by step.
A monitoring alert reports a RAID array in degraded state on a production file server, and a technician must assess risk and recover fault tolerance quickly.
- 01Check array statuscat /proc/mdstat shows md0 as active but with [UU_] instead of [UUU], meaning one of three drives has dropped out of the array.
- 02Get detail on the arraymdadm --detail /dev/md0 confirms one drive is marked as faulty and the array state is clean, degraded, meaning it still serves data but has no remaining fault tolerance.
- 03Identify the failed driveThe detail output lists /dev/sdc as the failed device, which is cross-referenced against the physical bay label in the chassis.
- 04Remove the failed drive from the arraymdadm /dev/md0 --remove /dev/sdc cleanly detaches the failed member so it can be physically replaced.
- 05Replace the physical driveThe failed drive is swapped for a new one of equal or greater capacity during a scheduled maintenance window.
- 06Add the replacement to the arraymdadm /dev/md0 --add /dev/sdc begins the rebuild process, recalculating parity across all three drives.
- 07Monitor the rebuildwatch cat /proc/mdstat shows a recovery percentage climbing until the array returns to [UUU], fully redundant again.
- 08OutcomeThe array is restored to full fault tolerance before a second drive failure could have caused actual data loss.
Outcome: Prompt identification and replacement closed the window of vulnerability during which the array could not survive a second failure.
Server hardware and storage array reference
Worth keeping at hand while you work.
- RAID 0
- Striping across drives for performance, no redundancy; any single drive failure loses all data.
- RAID 1
- Mirroring across two drives; survives one drive failure at 50 percent usable capacity.
- RAID 5
- Striping with single parity; survives one drive failure, usable capacity is total minus one drive.
- RAID 6
- Striping with double parity; survives two simultaneous drive failures, usable capacity is total minus two drives.
- RAID 10
- Mirrored pairs striped together; good performance and fault tolerance at 50 percent usable capacity.
- mdadm --detail /dev/mdX
- Reports array state, member disks, and whether the array is degraded.
- /proc/mdstat
- Quick view of all software RAID arrays and their current sync or degraded state.
- Hot spare
- An idle drive that automatically joins an array to begin rebuilding as soon as a failure is detected.
- Degraded array
- An array still functioning after a drive failure but with reduced or zero remaining fault tolerance.
- Redundant power supply
- Protects against a single power supply failure, unrelated to RAID's data protection role.
- RAID is not backup
- RAID protects against drive failure only, not deletion, corruption, ransomware, or site loss.
- NIC teaming/bonding
- Combines multiple network interfaces for redundancy or throughput, analogous in purpose to RAID for storage.
Common misunderstandings
What most beginners get wrong here.
RAID 5 protects data the same way a backup does.
RAID only protects against a physical drive failure; it does nothing against accidental deletion, corruption, or ransomware, which require actual backups.
A degraded array is already data loss.
A degraded array still serves data correctly after one tolerated failure; the risk is that it has lost its remaining fault tolerance for a further failure.
More drives in a RAID 5 array always means better protection.
RAID 5 always tolerates exactly one drive failure regardless of array size, and larger arrays actually increase risk during a long rebuild window.
RAID 1 and RAID 0 offer similar protection since both use two or more drives.
RAID 0 has no redundancy at all and loses all data if any single drive fails, while RAID 1 mirrors data and survives one drive failure.
Once a hot spare joins an array, no further action is required.
The failed drive should still be physically replaced and a new hot spare added, since the array is now running without a spare in reserve.
Exam traps
How the question writers try to catch you out.
- Server+ style questions expect you to calculate usable capacity for RAID 5, RAID 6, and RAID 10 given a specific number and size of drives.
- A scenario describing an array that still serves data after one drive failure is testing recognition of the term degraded, not failed.
- Questions frequently test that RAID is not a substitute for backup, especially in scenarios involving accidental deletion or ransomware.
- RAID 6 versus RAID 5 questions hinge on whether the scenario requires surviving two simultaneous drive failures.
- Hot spare questions expect you to know it reduces rebuild delay, not that it eliminates the need to physically replace the failed drive.
Check yourself
Answer in your head first, then reveal. This is not scored.
Why is RAID not considered a backup strategy?
How many drive failures can RAID 6 tolerate at once?
What does a degraded RAID array indicate?
What is the usable capacity of a four-drive RAID 10 array using 2TB drives?
What role does a hot spare play when a drive fails?
Quick reference
A condensed summary of the lesson above, for revision.
What It Is
Server platforms provide redundant power supplies, hot-swap drives, ECC memory, hardware RAID or HBA storage controllers, redundant network paths, and out-of-band management such as iDRAC, iLO, or IPMI that works even when the OS is down. Shared storage uses SAN protocols such as iSCSI and Fibre Channel or NAS protocols such as NFS and SMB.
Why It Matters
Availability commitments depend on knowing what is actually redundant. Two power supplies on one circuit, or dual NICs into one switch, provide far less protection than the diagram suggests.
How It Works
- Management controllers run independently of the OS and expose console, sensors, and power control.
- RAID controllers present logical volumes and handle rebuilds transparently to the OS.
- Shared storage exports block or file resources over a dedicated network path.
Where You See It
- Data centre racks, virtualisation clusters, branch server rooms, and colocation environments.
Key Terms
- Out-of-band management
- A dedicated controller for remote console and power control.
- ECC memory
- Memory that detects and corrects single-bit errors.
- Hot swap
- Replacing a component without powering down.
- HBA
- Host bus adapter presenting raw disks or SAN connectivity.
- Single point of failure
- A component whose loss stops the service.
Examples
- Out-of-band management lets you view POST output and reboot a hung server from home.
- Two power supplies deliver real redundancy only when connected to independent feeds.
Common Problems
- Predictive drive failures
- Unnoticed failed PSU
- Firmware mismatch across a cluster
- Management interface exposed or unreachable
How It Fails
- A failed redundant component goes unnoticed until its partner fails too.
- Firmware drift causes unpredictable behaviour during cluster operations.
- An internet-exposed management controller is a direct route to full hardware control.
How to Troubleshoot
- Read the management controller's hardware log before opening the chassis.
- Check controller and drive status before assuming an OS problem.
- Compare firmware versions across identical nodes when behaviour differs.
Practical Knowledge
- Alert on redundant component failure, not only on total outage.
- Keep management interfaces on an isolated, access-controlled network.
Exam Coverage
- Server form factors and components
- Redundancy and availability features
- Out-of-band management and storage connectivity
Interview Questions
- Why is dual power supply insufficient on its own?
- How would you diagnose a server that does not boot and has no console attached?
Watch and read
Verified official and reputable sources for this topic. Links open in a new tab.
Video training
Professor Messer video channel — general CompTIA training (no dedicated CompTIA Server+ course)
Professor Messer
WatchVideoFree
Lesson notes and bookmark
Notes and bookmarks for this lesson, saved with everything else you have marked.
No notes on this item yet.
Learning progress
0% across six evidence areas. Reading alone does not change progress.
Prerequisites
Next steps
- 01Locate the out-of-band management address and health log of a server you can access.
- 02Draw one production service and mark every remaining single point of failure.