To ensure that objects are well-protected, the HCP S Series Node uses erasure coding. With erasure coding, data is encoded and broken into multiple chunks that are then stored across multiple locations. Additionally, parity chunks generated from the data chunks are stored across multiple locations. Each data or parity chunk is stored in a different location.
When erasure-coding the data for an object, the S Series Node creates 20 data chunks and 6 parity chunks and stores each chunk on a different drive.
The S Series Node can reconstruct the data for an object from any combination of 20 of the data and parity chunks for that object. This ability means that the object is protected as long as at least 21 of the 26 drives used to store the data and parity chunks are available. If only 20 of the drives are available, the S Series Node can reconstruct the object data, but that data is no longer protected. If fewer than 20 of the drives are available, the S Series Node cannot reconstruct the object data, and the object is considered irreparable.