[BBLISA] Moving from RAID 0 to LVM RAID?

[Edward Ned Harvey]

Ok, let's step back once here.  I'm not trying to say that HW raid is right
for everyone in every situation.  I'm not trying to say SW raid is bad.
There are pros & cons.

With HW raid, you can safely enable write caching (assuming you have battery
backup) and this costs extra, but it greatly improves performance.  Dell
PERC ships with battery, and write caching enabled by default.  I have
measured up to 60x faster performance with the write cache enabled (40mins
vs 40secs to mkfs on 1.2Tb) but more normally 4x-5x faster performance, for
"normal" sized files, benchmarked with iozone.

With SW raid, you have the cheaper solution.  Also in the case of ZFS, it
can be more flexible than the options otherwise available in the HW raid -
PERC for example, only does RAID 0, 1, and 5.  While ZFS and others can
present more options to the sys admin.  

If we're talking high performance, high reliability computing:  I recommend
HW raid, BBU enabled, cache enabled, hotswappable drives.   Implement
redundancy either in HW or SW.

If we're talking simply protection from data loss:  I recommend any raid as
long as it provides redundancy (not raid 0) and has the ability to notify
the admin when a disk goes bad.