CiscoDNALab10 - Brocade DCX offers no protection against corrupted frames

[jtarrio75]

What a bunch of crap. This only demonstrates Cisco has no idea how storage networks work, or what the difference between store-and-forward and cut-through switching is. The only reason Cisco can drop the frames at the ingress port is because they store the whole frame before retransmitting it (store-and-forward switching vs. cut-through switching), resulting in a much higher switching latency (an order of magnitude). Frames are dropped by end devices at the hardware level and have no app impact.

[msdetrick]

This test is fundamentally flawed. An appropriate test would be how the directors abide by the ANSI/INCITS standard. The standard speaks to only marking FC frames by altering the EOF indicating the frame is corrupted. They had a couple things in mind when they decided on this. One was to enable faster switching using cut-through techniques beneficial to SCSI communications. Two, the end device needs to know that a portion of the sequence resulted in an error for proper error recover.

[yuser9876]

Excellent video, very informative. Thanks Chris!