Garrett D'Amore writes: > This could create problems such that Nico's proposed adaptive tuning > might not actually *eliminate* the performance bottleneck that Mike was > trying to eliminate for the problematic nxge devices in question.
I think it forces Nico's solution to use LLDP, because avoiding the toxic configurations means making sure that your chosen lower MRU is 'safe' on the network. Or it forces us to abandon that solution and require explicit configuration everywhere. I'd been thinking of it as merely a "sweet spot" in the transmit frame size, and that's something that could be handled adaptively. If it's something that needs to be configured deliberately down in the hardware and that alters the receive buffers as well, then we're back to a feature that cannot be the default, cannot be made "automatic" in any simple way, and that inhibits interoperability. > All of this suggests (rather strongly) to me that the nxge perf. > bottleneck is a serious hardware problem, and *that* information needs +1 -- James Carlson, Solaris Networking <[email protected]> Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 _______________________________________________ networking-discuss mailing list [email protected]
