Redundant Array of Independent Drives

I'll be focusing on RAID 0, which is a configuration for high performance using multiple drives.

RAID is a configuration on a system which can abstractly allows a collection of drives to be treated logically as a single drive. The purpose of this is either to achieve redundancy for hardening a system and/or achieving performance.

• RAID 0 = striping, no fault tolerence but efficient performance
  • when writing with two physical drives backing a RAID 0 drive, effectively half of the data is written to one and half to the other
• note the raid controller is the interface that handles how RAID is implemented, and can be software or hardware level
• /dev in linux is a collection of files representing devices that our system has access to
  • see aws documentation on device naming schemes on ec2
  • sda1 is for root, sd[b-z], hd[a-z], xvd[a-z] are possible names for aws mounted drives
• mdadm is a linux utility that allow us to monitor and manage software RAID devices
  • cat /proc/mdstat shows status of all raids
  • this is the primary entry point for verifying if a raid volume is configured
  • "Personalities" tells you raid level
  • md_123 can be pronounced "multiple devices #123"
    • active tells us it's working, started is still configuring, inactive means broken
    • the next space delimited list is the devices used
• LVM (Logical Volume Manager) is a thin software layer atop hard disks which creates an abstraction for partitioning, managing, replacing and backing up hard drives
  • logical volumes can be created to include RAID functionality, where a LV's parts can be striped across multiple physical as in RAID 0
  • on the flip side, many logical volumes can map to a single physical volume, where different paths in a unix filesystem are the multiple logical volumes sitting atop a single physical volume