Abstract: Techniques related to flexible record definitions for semi-structured data in a relational database system are disclosed. The techniques include a machine readable medium, on which are stored instructions for accessing flexible record definitions for efficient database storage, comprising instructions that when executed cause a device to: receive a first request to create a first row in a database table, the first request indicating a first set of columns associated with the first row, determine that columns of the first set of columns are included in a second set of columns associated with the database table, select a first variant, from a set of variants, the first variant associated with a third set of columns, the third set of columns including at least the first set of columns, and write the first row to the database table based on the first variant and the third set of columns.

Abstract: According to some embodiments, so-called “NS-Tree” indexes may be used to calculate whether all (or a portion) of two independent tables are consistent. The NS-Tree indexes may be comprised of various elements, e.g., a synchronization time, a primary key, a hash of the complete record itself, and an aggregate value associated with each index entry (e.g., an accumulated XOR value). At any point in the index, an entry may possess the accumulated aggregate value of all key entries in the table up to that point. That aggregate value may be used to validate data consistency with another table(s) maintaining the same index. Due to the unique composition of the NS-Tree index, users can also validate two data sets within a ‘sub-range’ of the entire data set. According to other embodiments, NS-Tree indexes may also be applied to two different clusters of nodes by applying the NS-Trees at a ‘federated’ level.

Abstract: According to some embodiments, so-called “NS-Tree” indexes may be used to calculate whether all (or a portion) of two independent tables are consistent. The NS-Tree indexes may be comprised of various elements, e.g., a synchronization time, a primary key, a hash of the complete record itself, and an aggregate value associated with each index entry (e.g., an accumulated XOR value). At any point in the index, an entry may possess the accumulated aggregate value of all key entries in the table up to that point. That aggregate value may be used to validate data consistency with another table(s) maintaining the same index. Due to the unique composition of the NS-Tree index, users can also validate two data sets within a ‘sub-range’ of the entire data set. According to other embodiments, NS-Tree indexes may also be applied to two different clusters of nodes by applying the NS-Trees at a ‘federated’ level.

Abstract: Techniques related to flexible record definitions for semi-structured data in a relational database system are disclosed. The techniques include a machine readable medium, on which are stored instructions for accessing flexible record definitions for efficient database storage, comprising instructions that when executed cause a device to: receive a first request to create a first row in a database table, the first request indicating a first set of columns associated with the first row, determine that columns of the first set of columns are included in a second set of columns associated with the database table, select a first variant, from a set of variants, the first variant associated with a third set of columns, the third set of columns including at least the first set of columns, and write the first row to the database table based on the first variant and the third set of columns.