Using databases for both transactions and analysis

Abstract

Among other things, requests are received for tasks to be performed in a database. The tasks include tasks that will cause changes in data of the database and other tasks that must be executed on data in the database that is consistent time-wise. At least some of the data that will be changed by the change-causing tasks comprises data that must remain time-wise consistent for the other tasks. The other tasks are executed in the database while the change-causing tasks are also being executed in the database, without requiring a replication or reconfiguration of the database.

What is claimed is: 
     
         1 . A computer-implemented method comprising
 executing analytical tasks on data of a database that is expressed according to a schema that is natively suited to online transaction processing,   the analytical tasks being executed without requiring any transformation of the schema according to which the data is expressed.   
     
     
         2 . The method of  claim 1  in which the schema conforms to an expressive semantic model that is of an arbitrarily high level of expressiveness. 
     
     
         3 . The method of  claim 1  comprising executing the tasks on data located at two or more different nodes of the network. 
     
     
         4 . The method of  claim 1  comprising executing the tasks using two or more processes running at two or more different nodes connected by communication links. 
     
     
         5 . The method of  claim 1  in which at least a portion of the data is stored in transient memory. 
     
     
         6 . The computer-implemented method of  claim 4  in which executing the analytical tasks comprises accessing versions of the data that are time-wise consistent and stored in accordance with the schema of the database. 
     
     
         7 . A computer-implemented method comprising
 executing an analytical task using data of a database,   the data to be used by the analytical task being subject to changes made by transactional tasks to be performed after the beginning of the executing of the analytical task and before the completion of the executing of the analytical task, and   guaranteeing time-wise consistency of the data used by the analytical task in executing the database notwithstanding the changes made by the transactional tasks.   
     
     
         8 . The method of  claim 6  in which guaranteeing time-wise consistency comprises assuring access by the analytical task to time-wise consistent data stored persistently in the database at known locations. 
     
     
         9 . The method of  claim 6  comprising executing the tasks on data located at two or more different nodes of the network. 
     
     
         10 . The method of  claim 6  comprising executing the tasks using two or more processes running at two or more different nodes connected by communication links. 
     
     
         11 . A computer-implemented method comprising
 if at least part of a relatively shorter latency operation is in contention with at least part of a relatively longer latency operation for access to data stored persistently at a node of a distributed database, temporarily suspending the longer latency operation while at least the part of the shorter latency operation proceeds, the longer latency operation requiring access to the data in its state as of a particular moment in time,   if data in its state as of the particular moment in time is to be changed by the shorter latency operation while the longer latency operation is suspended, then storing the changed data as a new version persistently at the node, without making the unchanged data inaccessible to the longer latency operation, and   permitting the longer latency operation to resume and to access the unchanged data in its state as of the particular moment in time.   
     
     
         12 . The method of  claim 1  in which at least a portion of the data is stored in transient memory.