Ulrich, You're most probably right. Nonetheless, I still believe that *in theory*, taking disk geometry into account, you can improve speed. However, the gain would be marginal and at the loss of CPU. Anyway, it's not really my field of expertise, so I actually shouldn't be having this discussion with you ;).
The problem is indeed with the software scheduler. Due to misaligment, one page on the initiator would require two pages on the target. Kind regards, Eric On Dec 22, 5:39 pm, "Ulrich Windl" <ulrich.wi...@rz.uni-regensburg.de> wrote: > On 22 Dec 2008 at 5:11, Eric wrote: > > > Besides that, I don't think I totally agree on your statement. Letting > > the IO scheduler taking the actual disk geometry into account, should > > give more performance. However, the fdisk default values are probably > > standard for disks these days. > > The disk scheduler can safely assume that contiguous nearby "forward" requests > perform better than small scattered "backwards" requests. All other is mostly > speculative IMHO. You would have to know about the heads position relative in > time, track skew, sector skew, etc. Today's disk have cache big enough for > whole > tracks, so the scheduler shouldn't really care about sectors. Also the > scheduler > really cannot know on which physical cylinder (or track, or head) a specific > block > will actually reside. > > However when we are talking about problems in the scheduler _software_ you > may be > right. I didn't look into it for years ;-) > > Regards, > Ulrich --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "open-iscsi" group. To post to this group, send email to email@example.com To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/open-iscsi -~----------~----~----~----~------~----~------~--~---