We have very bad read times when using default rsize/wsize. There was some discussion about using smaller values. Can someone summarize why and what size to use? I thought it was a router or switch bug that was to be fixed.

"smaller" (8k usually sufficient). The problem as I've seen it has been buffers getting overrun on cat6500's (and others, I'm sure), though enabling flowcontrol should do the trick.

All of this is in the context of gige, and you'll (hopefully) see drop counters increasing on the switch side of the filer's interface. I

..kg..