If the number of disks is not evenly divisible by the number of replicas you **reduce** your redundancy.

For example, two near copies with 5 disks. Should be better than with 4 disks, right? We have more disks, so less chance of failing?

sda1 | sdb1 | sdc1 | sde1 | sdf1 |

0 | 0 | 1 | 1 | 2 |

2 | 3 | 3 | 4 | 4 |

5 | 5 | 6 | 6 | 7 |

7 | 8 | 8 | 9 | 9 |

10 | 10 | 11 | 11 | 12 |

Imagine sda1 dies. Now we've lost one copy of block 0, so if sdb1 dies we're in trouble. But we've also lost one copy of block 2, so if sdf1 dies we're dead too. Ow. If we'd stuck to four disks, or increased to six disks we would have a raid that would survive all but failures of "adjacent" disks.

All questions to john@AtlanTech.com