Archives of the TeradataForum

Message Posted: Thu, 27 Oct 2005 @ 09:12:22 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

We used to use Raid-5 on 5150s but the write I/O was much slower than Raid-1 so we never seriously considered it on the 5350/5400 systems. The problem is the extra reads needed for parity update:

Raid-1: Write Primary, Write Mirror

Raid-5: Write Primary, Read Secondary#1, Read Secondary#2, Compute

Parity, Write Parity

Multiply by two for fallback.

The storage gain is really tempting, I know, but I'd recommend against it.

/dave hough

	<Prev	Next>		<<First	<Prev	Next>	Last>>