| CPC G06F 16/278 (2019.01) [G06F 11/2028 (2013.01); G06F 16/2379 (2019.01)] | 27 Claims |

|
1. A method for configuration of a super region in a database, comprising:
executing operations on one or more processors of one or more of a plurality of servers, the operations comprising:
receiving, from a client computing device, a selection of a super region comprising at least three database regions of a plurality of database regions, wherein the plurality of servers are geographically located within the plurality of database regions, wherein each of the at least three database regions comprises at least one of the plurality of servers;
providing, based on a region survivability goal, at least three voting replicas of a data partition stored by the plurality of servers in the at least three database regions, wherein at least one voting replica of the at least three voting replicas is stored within a first database region of the at least three database regions, wherein the at least three voting replicas are configured to participate in a consensus protocol to commit write operations to the data partition; and
based on a failure of the first database region that causes removal of the at least one voting replica from the first database region, causing storage of the at least one voting replica of the failed first database region (i) within the super region and (ii) external to the failed first database region, wherein before the failure of the first database region:
the at least three database regions each store less than a majority number of the at least three voting replicas;
the at least three database regions each store at least one of the at least three voting replicas;
at least one database region of the at least three database regions stores a threshold number of the at least three voting replicas; and
the threshold number is the majority number minus one.
|