For provisioning SAN volumes to high performance OLTP databases, which of the following describes a best practice?

Separating redo log volumes from data volumes onto different physical disks (or RAID groups/LUNs backed by different spindles) is a critical best practice for OLTP databases to prevent I/O contention, as redo logs have highly sequential write patterns while data volumes often have more random I/O.

Provisioning volumes for each database on separate fabrics is primarily for high availability and redundancy, not specifically for optimizing I/O performance of different database components like redo logs versus data.

While separating indexes, redo, and archive logs onto different storage arrays can provide maximum isolation and performance, it is a very expensive and often unnecessary practice for all but the most extreme performance requirements; separating onto different disks (as in option A) is a more common and effective best practice.

Creating redo logs from each database on the same disks as other redo logs or even data volumes would exacerbate I/O contention, which is contrary to best practices for high-performance databases.