SK0-005
Free SK0-005 practice questions
10 questions from our SK0-005 practice tests, each with the correct answer and a written explanation. Read them with the answers shown, or switch to quiz mode and answer first.
Question 1 of 10, Server Hardware Installation and Management
A RAID 5 array of six 4 TB drives is used for a write-heavy transaction log workload, and write latency is unacceptable. The administrator can rebuild the array with the same drives. Which change would MOST improve write performance while keeping fault tolerance?
- ARebuild as RAID 6 to spread writes across two parity drives
- BRebuild as RAID 5 with a larger stripe size of 1 MB per disk
- CRebuild as RAID 10 to remove the parity calculation penalty Correct answer
- DRebuild as RAID 0 and rely on nightly backups for protection
Answer and explanation: RAID 10 has no parity calculation, so each logical write costs two physical writes instead of the read-modify-write cycle of parity RAID, which greatly improves small random writes while still tolerating drive failures. RAID 6 adds a second parity block and makes the write penalty worse. RAID 0 is fast but removes fault tolerance, and a larger stripe size does not eliminate the parity write penalty for small log writes.
Question 2 of 10, Server Hardware Installation and Management. Choose 2.
A storage engineer is designing a Fibre Channel SAN for a database cluster that must have no single point of failure between the hosts and the array. Which TWO design elements are required? (Choose two.)
- AOne dual-port HBA per host with both ports to the same switch
- BA single high-port-count FC switch with redundant power supplies
- CTwo HBAs per host, each connected to a separate fabric switch Correct answer
- DMultipath I/O software on each host to manage the redundant paths Correct answer
- EHard zoning that places every host in one shared zone
Answer and explanation: Separate HBAs cabled to two independent fabrics remove the HBA and switch as single points of failure, and MPIO software is needed so the OS uses the redundant paths and fails over between them. A single switch, even with redundant power, is still one device, a dual-port HBA is still one card on one switch, and zoning design controls access rather than providing path redundancy.
Question 3 of 10, Server Administration. Choose 2.
A virtualization cluster must support live migration of VMs between hosts. Which TWO requirements are typically needed? (Choose two.)
- AA Type 2 hypervisor installed on top of each host OS
- BShared storage (or a storage migration capability) accessible by both hosts Correct answer
- CIdentical host names on both hypervisor hosts
- DA separate DHCP scope for each VM being migrated
- ECompatible CPU families or an enhanced compatibility mode between hosts Correct answer
Answer and explanation: Live migration requires the destination host to access the VM's disks (shared storage or combined storage migration) and compatible CPU features, often via EVC or a processor compatibility mode. Host names must be unique, VMs keep their existing IP configuration rather than needing per-VM scopes, and enterprise live migration is a Type 1 hypervisor feature.
Question 4 of 10, Server Administration
A server has four 10 GbE ports bundled with LACP into one port channel. A single large file copy between this server and one client reaches only about 10 Gb/s. What is the MOST likely reason?
- AThe client NIC must also have four ports before LACP will activate on the server side
- BLACP hashes each flow to a single member link, so one flow cannot exceed one link's speed Correct answer
- CThe port channel is misconfigured, because a correct LACP bond delivers 40 Gb/s to any single flow
- DThe switch must be set to round-robin mode before LACP will use all four links
Answer and explanation: LACP distributes traffic by hashing flow fields such as addresses and ports, so a single TCP flow always uses one member link and is limited to 10 Gb/s, while many flows spread across links. A correct bond does not give 40 Gb/s to one flow, switches do not use round-robin for LACP, and the client does not need matching ports for the server's bond to work.
Question 5 of 10, Server Administration
A hypervisor host has 64 physical cores, and the administrator has assigned a total of 192 vCPUs to its virtual machines. What is this practice called, and what is the main risk?
- ACPU overcommitment, which risks CPU ready time under load Correct answer
- BCPU hot-add, which risks the guest OS failing to detect the new vCPUs
- CCPU passthrough, which risks the host losing access to its own cores
- DCPU affinity, which risks VMs being locked to a single core permanently
Answer and explanation: Assigning more vCPUs than physical cores is overcommitment (here 3:1); it is common but causes CPU ready/wait time when many VMs demand CPU at once. Affinity pins VMs to specific cores, hot-add adds vCPUs to a running VM, and passthrough dedicates a physical device to a guest; none describe the ratio in the scenario.
Question 6 of 10, Security and Disaster Recovery
A backup administrator runs a full backup on Sunday and wants each weekday restore to require only the full backup plus ONE other backup set. Which backup type should run Monday through Saturday?
- ACopy backup
- BIncremental backup
- CDifferential backup Correct answer
- DSnapshot of the boot volume
Answer and explanation: A differential backup contains every change since the last full backup, so a restore needs only the full plus the latest differential. Incrementals capture changes since the previous backup of any type, so a restore needs the full plus every incremental. A boot volume snapshot does not cover all data, and a copy backup does not reset archive bits and is not a scheduling strategy for this goal.
Question 7 of 10, Security and Disaster Recovery
A retailer wants the lowest-cost recovery site that can still be operational within about two days. Equipment is installed but not kept current with production data. Which option BEST fits, and what must happen during a disaster?
- AA warm site, where backups are restored onto pre-installed equipment before use Correct answer
- BA cloud region with synchronous replication that fails over automatically
- CA hot site, where users simply connect to systems already running with live data
- DA cold site, where hardware is purchased and shipped before restoration begins
Answer and explanation: A warm site has hardware and connectivity in place but not current data, so recovery means restoring recent backups and configuring systems, typically within hours to days at moderate cost. A hot site is faster but the most expensive, a cold site needs equipment procured and would usually exceed two days, and synchronous replication with automatic failover is a hot-site design.
Question 8 of 10, Troubleshooting
During a RAID 5 rebuild on a 12-drive array of 16 TB SATA disks, a second drive reports unrecoverable read errors and the array fails. Which design change would MOST reduce the chance of this happening again?
- AIncrease the rebuild priority so the rebuild finishes before any other errors occur
- BReplace the drives with larger 20 TB disks so fewer drives are needed
- CUse RAID 6 or RAID 10 so the array survives an additional failure during rebuilds Correct answer
- DDisable patrol reads so the drives are not stressed between rebuilds
Answer and explanation: Large drives take a long time to rebuild and RAID 5 has no remaining redundancy during that time, so a single unrecoverable read error can fail the array; RAID 6 or RAID 10 tolerate an additional error. Higher rebuild priority shortens but does not remove the risk, larger drives lengthen rebuilds further, and patrol reads actually help find latent errors before a rebuild.
Question 9 of 10, Troubleshooting
A RAID 1 boot mirror was rebuilt after a drive replacement, but the server fails to boot when the original surviving drive is removed for testing. What is the MOST likely cause?
- AThe server's NIC firmware is out of date
- BThe boot loader exists only on the original drive Correct answer
- CThe new drive is larger than the original drive, so the controller rejects it as a boot device
- DThe RAID controller cache battery is discharged
Answer and explanation: With software or partially configured mirrors, the boot loader or EFI partition may not be copied to the replacement disk, so the server cannot boot from it alone; installing the boot loader on both disks fixes it. A larger replacement drive is normally fine, the cache battery affects write performance, and NIC firmware does not affect local boot.
Question 10 of 10, Troubleshooting
Servers on a new storage VLAN configured for jumbo frames can ping the iSCSI array with small packets, but iSCSI sessions hang during large transfers. What is the MOST likely cause?
- AAn MTU mismatch, with one hop still at a 1500-byte MTU Correct answer
- BA duplicate MAC address on the server's management NIC
- CAn expired TLS certificate on the array web interface used by the initiators
- DAn incorrect DNS record for the storage array's name
Answer and explanation: When jumbo frames are configured on some devices but not all, small packets pass but large frames are dropped, causing stalls; ping with don't-fragment and a large size (for example 8972 bytes) can locate the mismatch. DNS, a duplicate management MAC, or the web interface certificate would not cause size-dependent iSCSI failures.