System and method of data allocation providing increased reliability of storage

A distributed data-storage system for storing a block of data, the system comprising:

• a plurality of disks configured to store data;
• and
• a processor configured to:
• identify a list of the plurality of disks,
• select a disk from the list of the plurality of disks based on a respective identified failure domain of the selected disk and add the selected disk to a copyset comprising a candidate subset of the disks designated for allocation of different chunks of the block of data;
• continuously select additional disks from the list of the plurality of disks based on a respective identified failure domain and add the additional disks to the copyset until the copyset contains a predetermined amount of allocated disks from the distributed data-storage system,
• repeat the selecting the disk and the additional disks to allocate a plurality of subsets of disks for a plurality of copysets, respectively, such that is no more than one disk belonging to a particular failure domain in any of the plurality of copysets, generate a plurality of data chunks from the block of data, wherein an amount of data chunks is equal to the predetermined amount of allocated disks, wherein at least one data chunk includes erasure encoding for the block of data,
• select a copyset from the plurality of copysets;
• and
• randomly distribute the data chunks onto disks of the selected copyset.

A non-transitory computer readable medium storing computer executable instructions for storing a block of data in a distributed data-storage system, including instructions for:

• identifying, by a processor, a list of a plurality of disks in the distributed data-storage system;
• selecting, by the processor, a disk from the list of the plurality of disks based on a respective identified failure domain of the selected disk and adding the selected disk to a copyset comprising a candidate subset of the disks designated for allocation of different chunks of the block of data;
• continuously performing, by the processor, the step of selecting a disk based on a respective failure domain and adding the disk to the copyset until the copyset contains a predetermined amount of allocated disks from the distributed data-storage system;
• repeating, by the processor, the selecting and continuously performing steps to allocate a plurality of subsets of disks for a plurality of copysets, respectively, such that is no more than one disk belonging to a particular failure domain in any of the plurality of copysets;
• generating, by the processor, a plurality of data chunks divided from the block of data, wherein an amount of data chunks is equal to the predetermined amount of allocated disks, wherein at least one data chunk includes erasure encoding for the block of data;
• selecting a copyset from the plurality of copysets;
• and
• randomly distributing, by the processor, the data chunks onto disks of the selected copyset.