Method and apparatus for efficient aggregate computation over data streams

1. A method comprising:

determining, using a processor, a set of aggregate queries to be executed on a data stream, the set of aggregate queries comprising queries that perform respective sets of aggregation operations on respective sets of attribute values over respective time intervals;

generating, using the processor, at least one intermediate aggregate query for a subset of the set of aggregate queries, said at least one intermediate aggregate query combining a subset of aggregation operations for the subset of aggregate queries and a subset of attribute values;

executing, using the processor, said at least one intermediate aggregate query to generate pre-aggregated data from the data stream for the subset of aggregate queries; and

executing, using the processor, the subset of aggregate queries on the pre-aggregated data subsequent to executing said at least one intermediate aggregate query;

wherein each of at least two aggregate queries in the subset of aggregate queries comprises:

a number of group-by attributes on which aggregation is performed; and

a time interval over which aggregation is performed.

2. The method of claim 1 , wherein generating said at least one intermediate aggregate query further comprises determining that said at least one intermediate aggregate query reduces a computational cost of executing the set of aggregate queries to be executed on the data stream.

3. The method of claim 1 , wherein said at least one intermediate aggregate query comprises a number of group-by attributes, the number of group-by attributes in said at least one intermediate aggregate query being less than a sum of the numbers of group-by attributes in the subset of aggregate queries.

4. The method of claim 3 , wherein generating said at least one intermediate aggregate query further comprises determining that

S

<

N

*

(

X

-

Y

)

X

where N is a given input size of tuples in the data stream, S is the output size of tuples of said at least one intermediate aggregate query, X is the sum of the numbers of group-by attributes for aggregate queries in the subset of aggregate queries and Y is the number of group-by attributes in said at least one intermediate aggregate query.

5. The method of claim 1 , further comprising subjecting at least one of the aggregate queries in the subset of aggregate queries to a respective set of attribute filters specifying respective attribute range conditions for respective sets of attribute values associated with the at least one of the aggregate queries.

6. The method of claim 5 , wherein said at least one intermediate aggregate query is generated by combining respective attribute filters of two or more of the subset of aggregate queries to form a single attribute filter usable to pre-filter pre-aggregated data input to the two or more aggregate queries.

7. The method of claim 1 , wherein the data stream comprises network traffic records.

8. The method of claim 1 , wherein the data stream comprises Internet Protocol flow records.

9. The method of claim 1 , wherein the data stream comprises at least one of: sensor node readings; call detail records in a telecommunications network; retail transaction records; and one or more financial tickers.

10. An article of manufacture comprising a processor-readable non-transitory storage medium storing one or more instructions which, when executed by a processor, configure the processor to:

determine a set of aggregate queries to be executed on a data stream, the set of aggregate queries comprising queries that perform respective sets of aggregation operations on respective sets of attribute values over respective time intervals;

generate at least one intermediate aggregate query for a subset of the set of aggregate queries, said at least one intermediate aggregate query combining a subset of aggregation operations for the subset of aggregate queries and a subset of attribute values;

execute said at least one intermediate aggregate query to generate pre-aggregated data from the data stream for the subset of queries; and

execute the subset of aggregate queries on the pre-aggregated data subsequent to executing said at least one intermediate aggregate query;

wherein each of at least two aggregate queries in the subset of aggregate queries comprises:

a number of group-by attributes on which aggregation is performed; and

a time interval over which aggregation is performed.

11. The article of manufacture of claim 10 , wherein generating said at least one intermediate aggregate query further comprises determining that said at least one intermediate aggregate query reduces a computational cost of executing the set of aggregate queries to be executed on the data stream.

12. The article of manufacture of claim 10 , wherein said at least one intermediate aggregate query comprises a number of group-by attributes, the number of group-by attributes in said at least one intermediate aggregate query being less than a sum of the numbers of group-by attributes in the subset of aggregate queries.

13. The article of manufacture of claim 10 , wherein the one or more instructions, when executed by a processor, further configure the processor to subject at least one of the aggregate queries in the subset of aggregate queries to a respective set of attribute filters specifying respective attribute range conditions for respective sets of attribute values associated with the at least one of the aggregate queries.

14. The article of manufacture of claim 13 , wherein said at least one intermediate aggregate query is generated by combining respective attribute filters of two or more of the subset of aggregate queries to form a single attribute filter usable to pre-filter pre-aggregated data input to the two or more aggregate queries.

15. Apparatus, comprising:

a memory; and

a processor coupled to the memory and configured to:

determine a set of aggregate queries to be executed on a data stream, the set of aggregate queries comprising queries that perform respective sets of aggregation operations on respective sets of attribute values over respective time intervals;

generate at least one intermediate aggregate query for a subset of the set of aggregate queries, said at least one intermediate aggregate query combining a subset of aggregation operations for the subset of aggregate queries and a subset of attribute values;

execute said at least one intermediate aggregate query to generate pre-aggregated data from the data stream for the subset of queries; and

execute the subset of aggregate queries on the pre-aggregated data subsequent to executing said at least one intermediate aggregate query;

wherein each of at least two aggregate queries in the subset of aggregate queries comprises:

a number of group-by attributes on which aggregation is performed; and

a time interval over which aggregation is performed.

16. The apparatus of claim 15 , wherein the processor is configured to generate said at least one intermediate aggregate query by determining that said at least one intermediate aggregate query reduces a computational cost of executing the set of aggregate queries to be executed on the data stream.

17. The apparatus of claim 15 , wherein said at least one intermediate aggregate query comprises a number of group-by attributes, the number of group-by attributes in said at least one intermediate aggregate query being less than a sum of the numbers of group-by attributes in the subset of aggregate queries.

18. The apparatus of claim 15 , wherein the processor is further configured to subject at least one of the aggregate queries in the subset of aggregate queries to a respective set of attribute filters specifying respective attribute range conditions for respective sets of attribute values associated with the at least one of the aggregate queries.

19. The apparatus of claim 18 , wherein said at least one intermediate aggregate query is generated by combining respective attribute filters of two or more of the subset of aggregate queries to form a single attribute filter usable to pre-filter pre-aggregated data input to the two or more aggregate queries.

20. The apparatus of claim 15 , wherein the data stream comprises at least one of: network traffic records; sensor node readings; call detail records in a telecommunications network; retail transaction records; and one or more financial tickers.