views:

188

answers:

2

When configuring a database server with mechanical drives, mirroring hard drives or using RAID5 for redundancy in case of drive failure is the norm. Is this redundancy also required when using solid state disks or is the "mechanical" reliability of SSDs (write fatigue not withstanding) such that drive redundancy is unnecessary? This is of course assuming normal off-system backup routines and so forth.

There's a good discussion on SSDs and database scaling here.

+1  A: 

Yes, with any piece of electronics it may die on you suddenly. The MTBF or reliability of a SSD may be better than a HDD, but it is not 100% perfect.

So you have to weight the risk of failure against the value of the data or the cost of restoring from backups (remembering that backups don't tend to be continual so you'd lose a little data after a crash).

gbjbaanb
A: 

Just remember what they said about the Titanic.

Any piece of electrical hardware can die without obvious cause, just like that. You still need RAID to best prevent data loss.

Oli