A customer has identified a performance bottleneck with Online Transaction Processing (OLTP) production host LUNs. The performance improvement plan includes a change to the Hitachi Dynamic Link…

Question

Options

• ALeast I/O
• BLeast block
• CExtended Round Robin
• DExtended Least block

• A
  72% (23)
• B
  6% (2)
• C
  3% (1)
• D
  19% (6)

Explanation

Least I/O is correct because OLTP workloads generate high volumes of small, random I/O operations, and this algorithm dynamically routes each new request to the path with the fewest currently outstanding I/Os - preventing queue buildup on any single path and directly addressing the performance bottleneck.

Extended Round Robin (C) is still a static rotation scheme that ignores actual path load; it improves on basic Round Robin slightly but cannot adapt to real-time congestion, making it unsuitable for a bottleneck fix. Least Block (B) weighs the volume of data (bytes/blocks) in flight rather than the count of I/Os - this is better suited to mixed or large sequential workloads where I/O sizes vary widely, not for OLTP where I/Os are uniformly small. Extended Least Block (D) is not a standard HDLM algorithm option and is included as a plausible-sounding distractor.

Memory tip: Match the metric to the workload - OLTP is about transaction count (many small I/Os), so use the algorithm that counts I/Os in flight: Least I/O. If the workload were large sequential transfers (like backup or DSS), you'd think in blocks instead.

No community discussion yet for this question.