Handling Silent Relations In A Data Stream Management System

• US 20090100029A1
• Filed: 10/16/2007
• Published: 04/16/2009
• Est. Priority Date: 10/16/2007
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Jain, Namit, Srinivasan, Anand, Mishra, Shailendra Kumar
• Granted Patent: US 7,979,420 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2474 Sequence data queries, e.g. querying versioned data

Abstract

A new continuous query to a data stream management system (DSMS) may use a silent relation whose source does not provide any heartbeats. During execution of any given operator, the DSMS processes data from one input (“active input”) using its time stamp as if the same time stamp is received from another input whose relation has fallen silent (“empty input”), if one or more predetermined conditions are met. One predetermined condition is that the empty input'"'"'s operator has all its own inputs silent. And another predetermined condition is that a maximum time stamp that is currently present among all the inputs of the empty input'"'"'s operator is equal to a time stamp that was last received from these same inputs. When both conditions are met, previously received data from the empty input is time stamped with the same time stamp as the most recently received data from the active input.

Claims

1. A method, implemented in a computer, to manage a plurality of streams of data, and a plurality of relations that supply data infrequently relative to said plurality of streams, the method comprising:

awakening an operator to process said data, said operator comprising a plurality of inputs;

checking if any input in said plurality of inputs of said operator has no data;

based on outcome of said checking, assuming receipt from said any input, of a null message with a time stamp same as a latest message comprising said data, from among all inputs in said plurality of inputs if a group of predetermined conditions are met; and

outputting a stream from said computer, said stream being generated based at least partially on said operator processing said data using said null message.

2. The method of claim 1 wherein:

at least one predetermined condition in said group is whether all inputs, of an operator at said any input, have no data.

3. The method of claim 1 wherein:

at least one predetermined condition in said group is whether a maximum time stamp among all inputs, of an operator at said any input, is equal to a time stamp that was last received from said all inputs.

4. A method, implemented in a computer, to manage a plurality of streams of data, the method comprising:

receiving a new continuous query to be executed on at least a relation in a global plan to execute a plurality of queries on at least said plurality of streams;

creating for at least a first operator in the new continuous query, a list of operators in said global plan, wherein said operators supply data on the relation, and said data is supplied infrequently relative to said plurality of streams;

altering said global plan to include the new continuous query in addition to said plurality of continuous queries, thereby to result in an altered global plan;

wherein the altered global plan treats data received by a third operator at a first input not on said list as if a first time stamp is received at a second input of said third operator from a second operator on said list, said first time stamp being associated with the data from the first operator if a group of predetermined conditions are met;

wherein at least one predetermined condition in said group is if the second operator has no data to supply; and

storing the altered global plan in a memory of said computer.

5. The method of claim 4 wherein:

still another predetermined condition in said group comprises checking if a most-recently received time stamp of the second operator is equal to time stamps of all other inputs of the third operator.

6. The method of claim 4 wherein:

the relation is represented incrementally in the global plan, as a plurality of tuples with each tuple being time stamped; and

all tuples of the relation having a time stamp older than a current time in said computer.

7. A computer-readable storage medium encoded with instructions, said instructions comprising:

instructions to receive a new continuous query to be executed on at least a relation in a global plan to execute a plurality of queries on at least said plurality of streams;

instructions to create for at least a first operator in the new continuous query, a list of operators in said global plan, wherein said operators supply data on the relation, and said data is supplied infrequently relative to said plurality of streams; and

instructions to alter said global plan to include the new continuous query in addition to said plurality of continuous queries, thereby to result in an altered global plan;

wherein the altered global plan treats data received by a third operator at a first input not on said list as if a first time stamp is received at a second input of said third operator from a second operator on said list, said first time stamp being associated with the data from the first operator if a group of predetermined conditions are met;

wherein at least one predetermined condition in said group is if the second operator has no data to supply.