Preston: Erasure code use provides basic object-level redundancyDate: Jan 21, 2014
In an object-based storage system, erasure coding provides redundancy in case of data loss or corruption. In this Storage Decisions presentation, W. Curtis Preston, the founder of TruthInIT and BackupCentral.com, discusses how erasure code use protects data by dispersing it across a wide area network (WAN).
"The beauty of this is … I can lose all kinds of different combinations of disk drives at the same time and still function, I'm functioning at the same speed," said Preston.
WhatIs.com defines erasure coding as a method in which data is broken into fragments and stored in different locations as a means of data protection.
Preston noted that an object is a piece of data that can be addressed and manipulated in individual sections, often with metadata that includes information or tags to identify it.
Preston said a "very simplistic" example is to take an object and divide it into nine segments and then calculate three additional parity segments. All 12 segments are then distributed evenly through an array. In this arrangement, the object is still available as long as any nine of the 12 segments are available.
"The idea is that you can survive three simultaneous failures with no performance degradation," said Preston. "And if three isn't good enough for you, dial the knob … that's the beauty of an object-based storage system. This isn't like RAID 7 or anything like that."