An administrator is configuring Storage I/O Control. After enabling, the administrator notices high device latency and poor storage performance. Which two actions would likely reduce latency and…

Setting the congestion threshold to 15ms - lower than the default 30ms - causes SIOC to detect congestion and begin redistributing I/O credits sooner, preventing device latency from climbing to high values before throttling activates. Earlier intervention keeps average and peak latency lower for all competing workloads.

A 5ms congestion threshold is excessively aggressive - it would force SIOC to throttle storage I/O even under light, normal workloads where no real contention exists, artificially degrading throughput rather than improving it.

SIOC determines congestion by measuring per-datastore device latency, but when multiple datastores share the same physical spindles, I/O load on one datastore inflates the observed latency on others. Independent spindles per datastore ensure SIOC reads accurate, isolated latency values and can correctly enforce per-VM I/O limits.

Shared spindles between datastores corrupt SIOC's latency measurements because activity on one datastore raises the apparent latency seen by another, causing SIOC to throttle workloads incorrectly and worsening overall storage performance.