Had a server crash last night that changed my mind about raid 5

I've been running raid 5 arrays for years at this small office I maintain near downtown. Always thought it was the sweet spot between speed and redundancy. Well last night at 2am I get the call that their file server is screaming and two drives are showing as failed. Not one but two. I drove over there and sure enough the rebuild was toast because the second drive gave up during the initial degradation. Lost about 3 hours of work from that day because their backup was from the previous night. I'm switching that whole setup to raid 10 this weekend even if it costs more in drives. The downtime and data loss just isn't worth the storage savings for a business that has like 12 people accessing it. Has anyone else gotten burned by raid 5 like this or am I just unlucky with these Seagate drives?

2 comments

2 Comments

andrew_kelly15d ago

Yeah I read a blog post a while back from a storage engineer who said raid 5 is basically gambling on your drives dying at the worst possible time. He was saying that with these big drives we use now the rebuild time is so long that a second failure becomes way more likely than people realize. Your story pretty much proves his point right. I switched my home server to raid 10 last year after hearing too many horror stories like this and honestly the peace of mind is worth the extra cash for drives. Those two Seagates failing together is just brutal man, sorry you had to deal with that.

parkerkim15d ago

The phrase about "gambling on your drives dying at the worst possible time" really hits home after that night. I think it's worth noting though that the failure rates aren't totally random since Seagate had a bad batch of drives around that time as well. @andrew_kelly your storage engineer was right about the rebuild times on big drives making raid 5 risky, but the drive brand and model you pick also matters a ton for reliability.