Transaction prediction modeling method

摘要

The present invention provides database systems configured for creating and utilizing a transactions predictive model using a computer, the database system including: a database management system (DBMS) for handling transactions in the database system, the DBMS including a prediction component for generating the transactions predictive model; a database workload table for capturing commands associated with the transactions, the commands utilized for generating the transactions predictive model; database data for providing data in response to the transactions; and a prediction model database for storing the transactions predictive model, where the prediction component is configured for generating a generalized statement for each statement corresponding with the transactions, identifying a sequence of transactions of the transactions, creating a transaction class based on the sequence of transactions, identifying a current transaction, and then calculating probabilities between transactions, where each statement includes at least a database command.

主权项

1. A database system configured for creating and utilizing a plurality of transactions predictive models using a computer, the database system comprising:
a database management system (DBMS) for handling a plurality of database transactions in the database system, the DBMS including a prediction component for generating the plurality of transactions predictive models; a database workload table for capturing a plurality of commands associated with the plurality of database transactions, the plurality of commands utilized for generating the plurality of transactions predictive model; database data for providing data in response to the plurality of database transactions; and a prediction model database for storing the plurality of database transactions predictive models, wherein the prediction component is configured for
generating a generalized statement for each statement corresponding with the plurality of database transactions in the database workload table by replacing a variable parameter with a constant value, wherein the generalized statement is anonymized from the statement corresponding with the plurality of database transactions,identifying a sequence of transactions of the plurality of transactions, creating a plurality of transaction classes based on grouping the sequence of transactions having a same generalized statement,identifying a current database transaction, andcalculating transition probabilities of possible sequences between the plurality of transaction classes to generate the transactions prediction model, wherein each statement includes at least a database command, and the database system wherein the prediction component is further configured to receive the current database transaction, to identify one of the plurality of transaction classes corresponding with the current database transaction from one of the plurality of transactions prediction models, to predict a next transaction class based on a highest transition probability of the one of the plurality of transactions prediction models, and to pre-fetch data associated with the next transaction class.