RAID1 (simple mirroring) works better than raid 5. With the low price of drives today there is no reason NOT to mirror drives on a server. Using separate controllers (hosts) for primary and secondary drives makes it even more reliable.
Mirrored, not mirrored, it really wouldn't make much of a difference in my case. The backups are, as I said, excellent (and versioned), and the recovery procedure is quite simple. However, Raid or not, I can't expect my customers to be replacing system drives in machines that are under my warranty.
If you wind up with a corrupted sector (not unreadable, just corrupted) how do you tell which copy is good? The RAID5 (and especially the raidz2) has lots of checksumming so it can *tell* which sectors are bad and re-build that information from the data scattered over the other drives.
Do they also handle RAID5 as a built-in?
Sun's zfs will do mirroring or lots of other RAID options, and it is built-in. It will also allow building of drives of an insane size by combining the various RAID arrays. The 'z' in "zfs" stands for "zetabyte". :-)
I agree - and it's not just hard drives - or computer parts for all that matters. EVERYTHING today is made to sell at the lowest possible price because Americans, in particular, won't pay for quality. Because of that, nothing is made on this side of the pond anymore, factory workers have no work, so nobody is buying ANYTHING.
It is a bit ironic to hear "quality" discussed in this group with the frequent mention of Harbor Freight....
But to be more serious, nearly everything is sold, today, based on price. The idea of buying something because "it will last for ever" seems to be old fashioned, or at least very uncommon and as a result if a company can sell millions of units at the lowest cost and one or two "that will last forever" the long life models will disappear.
Years ago I worked in Indonesia and Caterpillar Tractor was being outsold by the cheaper Japanese equipment although Cat could prove that their equipment was cheaper over a, say 5 year, service life. Nobody cared. the only factor was sticker price.
"Somewhere a True Believer is training to kill you. He is training with minimum food or water,in austere conditions, day and night. The only thing clean on him is his weapon. He doesn't worry about what workout to do--- his rucksack weighs what it weighs, and he runs until the enemy stops chasing him. The True Believer doesn't care 'how hard it is'; he knows he either wins or he dies. He doesn't go home at 1700; he is home. He knows only the 'Cause.' Now, who wants to quit?"
NCOIC of the Special Forces Assessment and Selection Course in a welcome speech to new SF candidates
I've seen techs twist a drive quickly to break loose a stuck platter. Sometimes it worked. He said the bearing grease evaporated, condensed and glued the head down, or some such thing.
The stuck condition is often referred to as: sticktion. I don't know for sure, but I sorta doubt that it's due to bearing grease residue getting into the platter cavity.
A quick twisting motion is a "definite maybe" recovery method, at least for a short time, until it sticks again. As long as the drive is kept spinning, it will probably not stick again until it's shut down.
That's what I saw, immediately back it up onto a server or have IT clone it because it may not restart a second time. I was, um, discouraged from opening up networked company computers to play these games myself. However stand-alone test stations with custom hardware were all my problem.
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.