A large financial institution operates a dual-fabric Fibre Channel SAN with two separate MDS 9710 directors. Each fabric has multiple storage arrays and hundreds of hosts. The SAN is configured with VSANs to isolate different environments (production, development, backup). Recently, the backup VSAN has been experiencing slow performance during backup windows. Analysis shows that the ISLs between the directors in the backup VSAN are heavily utilized (near 100%) while other ISLs have spare capacity. The backup traffic consists of large sequential reads and writes. The SAN administrator has confirmed that there are no CRC errors or link issues. The backup VSAN uses a single 16 Gbps ISL. Which of the following is the best solution to improve backup performance?
Increases bandwidth and load balancing.
Why this answer
Adding an additional 16 Gbps ISL and configuring a port channel for the backup VSAN increases the available bandwidth for backup traffic, which consists of large sequential reads and writes that can fully utilize the link. Port channels provide load balancing across member links based on source/destination IDs, effectively distributing the backup traffic and reducing congestion on the single ISL. This directly addresses the near-100% utilization without introducing complexity or relying on other fabrics.
Exam trap
Cisco often tests the misconception that QoS or buffer tuning can solve bandwidth saturation issues, but the core problem here is insufficient aggregate bandwidth, which only adding physical links (via port channels) can resolve.
How to eliminate wrong answers
Option B is wrong because reducing buffer credits on the ISL would actually increase latency and potentially cause frame drops, worsening performance, especially for long-distance links; buffer credits are used to absorb link latency, not to reduce it. Option C is wrong because QoS prioritization does not increase available bandwidth; it only reorders traffic, and since backup traffic is already the only traffic on that VSAN, prioritizing it would have no effect on the high utilization. Option D is wrong because IVR (Inter-VSAN Routing) would route backup traffic through the other fabric, but that fabric's ISLs are also shared with other VSANs and may not have spare capacity; moreover, IVR introduces additional complexity and potential security risks without guaranteeing improved performance, and the problem is bandwidth scarcity within the backup VSAN itself.