diff --git a/data/mdadm.md b/data/mdadm.md
new file mode 100644
index 0000000..6362db3
--- /dev/null
+++ b/data/mdadm.md
@@ -0,0 +1,40 @@
+# RAID5
+
+You will need 4 disks and the `mdadm` package.
+The total size will be equal to the disks x 3, because one will be used for redundancy.
+
+> sudo mdadm --create --verbose /dev/*md127* --level=5 --raid-devices=*4* */dev/sdb /dev/sdc /dev/sdd /dev/sde*
+
+Note the variable parts:
+
+- The name of the device could be `/dev/md12` or whatever
+- The number of devices could be larger, but must be at least 4 for raid 5
+- We end by listing all devices in the new `md` device.
+
+Now look at how the raid status:
+
+> cat /proc/mdstat
+
+This will increase until the entire thing is fine.
+
+Check the health of your `mdadm` array:
+
+> sudo mdadm --detail /dev/md127
+
+You should see `State : clean`. If you see it is `degraded`, then a disk has broken.
+
+## Replacing a Disk
+
+First, prepare the disk with both a gpt partition table, and a partition:
+
+> sudo parted --script /dev/sdb mklabel gpt mkpart primary 1MiB -2048s
+
+Add the appropriate file system to it, e.g.:
+
+> sudo mkfs.ext4 /dev/sdb1
+
+Then finally, add the disk:
+
+> sudo mdadm --add /dev/md127 /dev/sdb1
+
+