Database systems architecture incorporating distributed log

A system comprising:

• at least one data processor;
• memory storing instructions which, when executed by the at least one data processor, result in operations comprising:
• receiving, from a client, a query specifying at least one database operation implicating data within a first database and within a second database, the first database comprising a first transaction manager and the second database comprising a second transaction manager;
• initiating a plurality of transactions responsive to the query to access the data within the first database and the second database;
• coordinating, using the first transaction manager and the second transaction manager, execution of the plurality of transactions;
• logging all changes specified by the at least one database operation in a distributed log in the first database including changes implicating data stored within the second database, the distributed log being accessible by both of the first transaction manager and the second transaction manager;
• and
• providing data responsive to the query to the client;
• wherein:
• streams of data are bijectively mapped to virtual files in persistence of the first database such that:
• if a virtual file is to be created, a lock is obtained to synchronize all file creations;
• a dedicated transaction is started that is independent from an outer database manipulation language (DML) transaction;
• the virtual file is created;
• entries are added to the virtual file during DML operations and during a precommit;
• an inner transaction is committed and made visible to all other concurrent transactions;
• and
• if an outer DML operation is rolled back, the created virtual file remains as it was committed in its dedicated transaction.

A non-transitory computer program product storing instructions which, when executed by at least one data processor, result in operations comprising:

• receiving, from a client, a query specifying at least one database operation implicating data within a first database and within a second database, the first database comprising a first transaction manager and the second database comprising a second transaction manager;
• initiating a plurality of transactions responsive to the query to access the data within the first database and the second database;
• coordinating, using the first transaction manager and the second transaction manager, execution of the plurality of transactions;
• logging all changes specified by the at least one database operation in a distributed log in the first database including changes implicating data stored within the second database, the distributed log being accessible by both of the first transaction manager and the second transaction manager;
• and
• providing data responsive to the query to the client;
• wherein:
• streams of data are bijectively mapped to virtual files in persistence of the first database such that:
• if a virtual file is to be created, a lock is obtained to synchronize all file creations;
• a dedicated transaction is started that is independent from an outer database manipulation language (DML) transaction;
• the virtual file is created;
• entries are added to the virtual file during DML operations and during a precommit;
• an inner transaction is committed and made visible to all other concurrent transactions;
• and
• if an outer DML operation is rolled back, the created virtual file remains as it was committed in its dedicated transaction.