On 04/22/11 05:55 AM, Dan Swartzendruber wrote: > Hmmm, looks like it might be realtek lossage. Crystaldiskmark just finished > the read phase. Getting about 56MB/sec, which isn't tremendous, but it > beats the snot out of the 33 or so the RT was generating. I then re-ran the > iSCSI crystaldiskmark test, and got about the same amount! e.g. cifs is > faster with the e1000 but iscsi slower. Not sure if there is some tuning > parameters for the e1000 card on the win7 that are dragging things down. > It's late and I think I've beat my head on this enough for tonight :) I was just reading DTrace book Chapter 5 and on the page of 387 and up (page 97 and up in Free to download 100 page sample PDF chapter from the book), I found an example of how to use Dtrace and other tools to track down performance problems, like one you explained.
Maybe is worth checking it out that Case study, in it, it fixed bandwidth problem with changing I/O size on the client from 1MB to 128KB. http://www.dtracebook.com/index.php/Main_Page http://dtracebook.com/images/dtbook_scripts.tar.gz Sample chapter: http://dtrace.org/blogs/brendan/files/2011/02/DTrace_Chapter_5_File_Systems.pdf It might also put DTrace in use on other problems too, that might be fixed by tweaking setting, segarding usage case and application needs. DTrace: http://www.solarisinternals.com/wiki/index.php/Solaris_Internals_and_Performance_FAQ http://hub.opensolaris.org/bin/view/Community+Group+dtrace/WebHome http://www.solarisinternals.com/wiki/index.php/DTrace_Topics_Intro http://wiki.genunix.org/wiki/index.php/DTrace_FAQ _______________________________________________ OpenIndiana-discuss mailing list [email protected] http://openindiana.org/mailman/listinfo/openindiana-discuss
