From: Pascal R. <pre...@is...> - 2005-05-03 15:03:44
|
Yeah - this is a problem, but you also have the difference between ext3 on one side and ntfs on the other. So the extrapolation between MS and Linux is really difficult to do there. But i will try to find a decent machine to re-run those. Pascal. >Hmmm, well those are some good tests, but unfortunately not very useful. How >do you extrapolate the difference in CPU? Your Win boxen has a CPU that is >at least 4x faster than the Linux boxes AND it has 4x the memory (not to >mention the architecture diff from a P4 and PIII). You just can't compare. >If you could run the same tests on the diff hw but run all the OS on each, >then that would be some good data. > >~kenny > > >Unless someone else has some better ideas. > >* Pascal Renauld <pre...@is...> [2005-05-03 10:33:28 -0400]: > > > >>Hi there, >> >>I have been doing some performance comparaisons between >>the different iSCSI initiators (Linux, Open-iSCSi and Microsoft) >>using the IET 0.4.7 and a commercial implementation on NetBSD. >>All the tests were conducted using Iometer with 100% sequential >>access and a serie of 5 runs (100% write, 80%write 20% read, >>50% write 50% read, 20% write 80% read, 100% read) for >>different transfer sizes (8KB to 128KB). >>The results are attached in the included Excel file. >>One of my question is why is the IET much slower in the 100% write >>runs compared to the NetBSD implementation? Thoughts? >>What do people think about the numbers? How can they be improved? >> >>Pascal. >> >> >> > > > > >------------------------------------------------------- >This SF.Net email is sponsored by: NEC IT Guy Games. >Get your fingers limbered up and give it your best shot. 4 great events, 4 >opportunities to win big! Highest score wins.NEC IT Guy Games. Play to >win an NEC 61 plasma display. Visit http://www.necitguy.com/?r=20 >_______________________________________________ >Iscsitarget-devel mailing list >Isc...@li... >https://lists.sourceforge.net/lists/listinfo/iscsitarget-devel > > > > > |