On Tue, Jun 15, 2010 at 11:14:38AM +0800, james wrote: > The stuff below is interesting and a reference, but this highlights my > favourite rant: Seagate's 'ATA more than an interface' says multiple disks in > a machine *will* result in a higher failure rate, maybe much higher.
Due to heat, or what? That paper seems to concern itself primarily with the differences between PS (personal storage) drives and ES (enterprise storage), in order to justify why the SCSI drives have so much higher cost per bit. The only mention I could see about multiple disks affecting failure rate was "A high density server rack with many disc drives grouped close together may experience much higher temperatures than a single drive mounted in a desktop computer". Nothing about whether multiple disks in a machine affect failure rate for any reason other than high temperature (which is usually controlled in server environments). > So raid is a less worse option than LVM. Heed the advice in slug talks about > backup (Sorry Sonia and Margurite, I don't remember who presented them) Yes. > It is possible, but not likely that *every* file on your disks is distributed > over all 3 disks, so worst cast is that you lost 1/3 of every file you have. Only if the Logical Volume is defined with striping (the -i argument to lvcreate). Rule #1 is always ... make backups. After that: - RAID1 can reduce the impact of a single-drive failure - RAID5 will increase the impact of failures - When combining multiple disks into a large Volume Group (VG), it is possible to create Logical Volumes within the VG so that they do not span physical devices. That way, if a disk dies (or 2, in a RAID1 setup) the entire VG contents will not be lost, only those filesystems on the failing devices. Hence it is a good idea to make multiple filesystems sized according to need. - Make multiple types of backups: backup to HDD (on a different server), offsite backup, Internet backup, incremental backups, DVD backups, external HDDs are dirt cheap these days. - Separate data according to importance and increase the redundancy level for the most important data. Data which is unimportant or can be recreated need not be backed up at all. Precious data might have multiple backups to onsite, offsite and write-once media. Nick. -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
