RAID & Erasure Coding

RAID stands for redundant array of independent disks. The incoming data is divided or replicated across multiple drives to deliver performance. This used to be the trusted protection available in the traditional storage array world until SDS came.

Advantages of RAID

Predictable disk performance based on disks

Disadvantages of RAID

High-capacity drive rebuilds can take days or weeks
Performance bottlenecks often

RAID 0 – Striping

This raid offers high performance for writes as data is spread across different disks & its spindles, however, has no protection for data against disk failures.

RAID 1 – Mirroring

This raid offers redundant copies of data and high-performance reads. Writes are slower as data needs to be mirrored to peer disks.

RAID 1+0 – Stripe of Mirrors

Data is striped across all mirrors and this offers high-performance reads and writes.

RAID 4 – Dedicated Parity

In this raid, one disk is dedicated to parity and this disk offers parity to the data disk should it fail.

RAID 5 – Distributed Parity

In this raid, parity is distributed across all disks so there is no requirement for a dedicated disk to store parity info.

RAID 6 – Dual Distributed Parity

This raid offers dual parity protection and parity info is spread across all disks.

Erasure Coding

Erasure coding is similar to RAID but on a per-object level. Data is broken down into fragments and encoded. Should any number of drives fail, all of the nodes in the cluster work in parallel to heal the situation as quickly as possible, so data is fully protected once again. The cluster can be configured to meet any requirements for uptime.

Advantages of Erasure Coding