A systems administrator is tasked with upgrading a database server's storage array, which currently uses a four-drive RAID 5 configuration. The primary goal of the upgrade is to significantly improve write performance for transaction-heavy database operations while maintaining fault tolerance for at least a single drive failure. The budget allows for a new four-drive array. Which RAID level should the administrator deploy to best meet these requirements?
The correct answer is RAID 10. This RAID level combines disk striping (like RAID 0 for performance) and disk mirroring (like RAID 1 for redundancy). For write-intensive applications like a transactional database, RAID 10 is superior to parity-based RAID levels because it does not have a parity calculation overhead, also known as a 'write penalty'. It also meets the requirement for single-drive fault tolerance.
RAID 6 is incorrect because it uses two parity blocks. This offers greater fault tolerance (up to two drive failures) but incurs an even higher write penalty than RAID 5, which would decrease, not improve, write performance.
RAID 5 is incorrect because it is the existing configuration that the administrator needs to improve upon. RAID 5 has a known write penalty due to its single parity calculation, making it less than ideal for the stated goal of significantly improving write performance.
RAID 0 is incorrect because, while it offers the highest write performance, it provides no fault tolerance. The failure of a single drive in a RAID 0 array results in the loss of all data, which violates the requirement to maintain at least single-drive fault tolerance.