Method for executing queries on streaming data using graphic processing units

A system for real time execution of SQL queries on at least one data stream using only HWA units, wherein the at least one data stream is injected into the one more HWA units, said system comprised of:

• a CPU unit comprised of an SQL analyzer module and decision module, wherein the SQL analyzer is programmed to receive an SQL query and identify said SQL query type;
• an HWA unit comprised of:
• a statistics analyzer module programmed for real time analyzing multiple data streams and calculating statistics coefficients and characteristics of at least one data stream, and, in case the calculated statistics coefficients and characteristics obey predefined rules, creating metadata based on statistical analysis;
• and
• an execution module programmed for using the created metadata for SQL execution based on SQL query identified type in case the calculated statistics coefficients and characteristics obey the predefined rules;
• wherein the real time statistical analysis includes at least one of:
• data distribution pattern in time, data points are independent probabilistically one to another, no correlation or no clustering between data units;
• wherein the statistical analysis based on which the metadata is created comprises a bloom filter;
• wherein the creation of metadata for the data stream for sort operation, includes, in case data is distributed with a predetermined degree of uniformity as determined from a data distribution pattern in time, applying brute force sorting operation on the data stream;
• and
• wherein the execution module is programmed to execute the SQL queries using said created metadata, such that the execution is based on SQL query identified type.

A method for real time execution of SQL queries on at least one data stream using HWA units, said method comprising the step of:

• receiving at least one data stream and injecting said data stream into the one more HWA units;
• receiving an SQL query and identifying said SQL query type;
• real time statistical analyzing at least one data stream, said statistical analysis including:
• calculating statistics coefficients and characteristics of the at least one data stream;
• operating at least one processor programmed, in case the calculated statistics coefficients and characteristics obey predefined rules, to create metadata based on statistical analysis;
• and, in case the calculated statistics coefficients and characteristics obey the predefined rules, to execute the SQL queries using said created metadata, such that the execution is based on SQL query identified type;
• wherein the steps of statistical data analysis and creation of meta data are performed by the HWA;
• wherein the real time statistical analysis includes at least one of:
• data distribution pattern in time, data points are independent probabilistically one to another, no correlation or no clustering between data units;
• wherein the creation of metadata for the data stream for sort operation, includes dynamically clustering data groups for sorting, including data elements within each cluster being optimized to fit onboard memory size of the HWA unit;
• and
• further comprising executing the SQL queries using said created metadata, such that the execution is based on SQL query identified type.

A System for real time execution of SQL queries on at least one data stream using only HWA units, wherein the at least one data stream is injected into the one more HWA units, said system comprised of:

• a CPU unit comprised of an SQL analyzer module and decision module, wherein the SQL analyzer is programmed to receive an SQL query and identify said SQL query type;
• an HWA unit comprised of:
• a statistics analyzer module programmed for real time analyzing multiple data streams and calculating statistics coefficients and characteristics of at least one data stream, and, in case the calculated statistics coefficients and characteristics obey predefined rules, creating metadata based on statistical analysis;
• an execution module programmed for using created metadata for SQL execution based on SQL query identified type in case the calculated statistics coefficients and characteristics obey the predefined rules;
• wherein the real time statistical analysis includes at least one of:
• data distribution pattern in time, data points are independent probabilistically one to another, no correlation or no clustering between data units;
• wherein the statistical analysis based on which the metadata is created comprises a bloom filter;
• wherein the creation of metadata for the data stream for sort operation, includes dynamically clustering data groups for sorting, including data elements within each cluster being optimized to fit onboard memory size of the HWA unit;
• and
• wherein the execution module is programmed to execute the SQL queries using said created metadata, such that the execution is based on SQL query identified type.