Your server has two existing hard disks, and you add a third 100-GB hard disk to speed up a client–server application that is slow when returning query results. To maximize read/write throughput for this workload, which RAID level should you choose?

Introduction / Context:
Disk I/O can bottleneck database-style client–server applications. RAID levels trade performance, capacity, and fault tolerance differently. When the explicit goal is the highest possible performance and no requirement for redundancy is stated, RAID 0 (striping) is the fastest option because it distributes reads and writes across multiple spindles without parity overhead.

Given Data / Assumptions:

• Additional 100-GB disk is added to improve performance.
• Primary concern is speed, not resiliency.
• Typical workload: many reads and writes associated with query execution and temp files.

Concept / Approach:

RAID 0 splits data into stripes written across disks concurrently, increasing aggregate throughput and IOPS. RAID 1 mirrors data for redundancy but provides at best modest write performance improvements and halves usable capacity. RAID 5 adds parity for fault tolerance but incurs a write penalty (read-modify-write), which slows small random writes common in query workloads.

Step-by-Step Solution:

Verification / Alternative check:

Benchmark before/after with representative queries to confirm lower latency and higher throughput on the striped volume.

Why Other Options Are Wrong:

RAID 1: Prioritizes availability; write performance gains are limited.

RAID 5: Write penalty harms performance under frequent updates.

RAID 2: Obsolete/not implemented in commodity controllers.

None: A clear performance-maximizing choice exists.

Common Pitfalls:

Ignoring that RAID 0 has no fault tolerance—if any disk fails, the volume is lost. Use backups and consider RAID 10 if both speed and redundancy are required.

Final Answer:

RAID 0 (striping)