It's just overhead, extra complexity with no benefit. There's no need for LVM in your scenario, so I'd eliminate that portion. I've just recently upgraded, but I ran my fileserver on an AMD 3000+ (1.8 Ghz single core processor) for years and could saturate a gigabit connection with it. Seeking for confirmation of my thinking or any other ideas/proposals/critics/anything for me to move forward.Īn i3 is massive overkill just for mdadm. ![]() Now I'm almost sure that i3 processor just can't afford syncing of 8TB of raw space in any SW RAID config normally. Simple partition on one of the disks - speeds ~ 600Mbps. Then tried 2 RAID1 arrays + LVM to make one volume because idea was that striping eats processor resource. IO speeds of the array like ~2Mbps, load averages above 4 while copying data. First version was RAID1+0 (I mean two RAID1 arrays and then RAID0 on top of those), then strightforward RAID10. Created software RAID (using Webmin, but checked mdadm config manually as there were some problems initially). First thing was to set up it as file server. Installed Ubuntu server 11.04, nice and smoothly. Asus P7F7-E WS supercomputer motherboard, INTEL CORE I3-560 3.33G processor, 4GB RAM, INTEL 510 SSD MLC 120GB for system disk and 4 HDDs for data: WD Caviar Green, 2TB, SATA/300, 64MB cache. I have built desktop HW intended for home file server. Now struggling for couple of month with software RAID performance issue. ![]() I'm not a total newbie, but not very experienced also :) Have enjoyed Ubuntu Desktop since 9th version.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |