I'm rebuilding an array and am debating on RAID 5 or RAID 6. Here's the requirements:
This server is very heavy on Writes, but not many Reads (it's a video recording server for our security cameras). The application requires a single volume and can't write to multiple logical drives. The drive will be full within 30 days and will then use a FIFO rotation scheme. The total capacity has to be in the 2.5TB range. I have (8) 450G drives to work with in an HP MSA60 chassis connected to a DL380 G4 running Windows 2003 R2.
Because my volume is over 2T, I'm using a GPT partition table vs MBR, but that shouldn't really matter in this scenario.
I could go with a single RAID 6 config with 2 hot spares, that will give me about 2.5T (6*450- 5%).
Or I could go with 2 RAID 5 arrays then span them in the OS to give me 2.5T (3*450 - 5% * 2).
What I'm really concerned about is the rebuild time if loose a drive. At 4G per hour, it will take almost 30 days to rebuild parity - probably more because of the sheer volume of data that is replaced every day. Is it better to have 2 spares in RAID 6, or smaller parity rebuild by using 2 RAID 5? I've never been a fan of spanning partitions, how does that play into this equation?