The paper [1] is for systems programmers, is quite accessible (modulo an error corrected in a follow-up 7 years later [2]) and explains a strategy to survive any number of disk failures by maintaining checksums using Galois Fields (which it also explains)

You may find it useful to also look at 'The mathematics of RAID-6' [3] explaining what the linux kernels RAID6 implementation does.

1: http://web.eecs.utk.edu/~plank/plank/papers/SPE-9-97.html

2: http://web.eecs.utk.edu/~plank/plank/papers/CS-03-504.html

3: https://www.kernel.org/pub/linux/kernel/people/hpa/raid6.pdf