Network traffic data summarization

1. An apparatus comprising:

a processor;

a memory on which is stored machine readable instructions executable by the processor to:

access network traffic data pertaining to data flows among nodes in a network;

partition the network traffic data into a plurality of windows;

for each of the plurality of windows, aggregate data flows between pairs of nodes;

compute a data distribution of each of the aggregated data flows;

select a summary structure for each of the aggregated data flows based on the computed data distributions of the aggregated data flows;

generate a summary of each of the aggregated data flows using the selected summary structures for the aggregated data flows;

store the generated summaries;

classify queries applied to a summary of an aggregated data flow; and

select a new summary structure for the aggregated data flow based on the classified queries.

2. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to partition the network traffic data into windows based on associated time stamps, on record counts, or on byte counts.

3. The apparatus of claim 1 , wherein the summary structure is at least one of:

a synopsis;

a data sample;

a histogram;

a wavelet;

a count;

a bit vector; or

a sketch.

4. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to aggregate the network traffic data at a broker prior to the partitioning of the network traffic data.

5. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to maintain the selected summary structure for each of the aggregated data flows until new network traffic data is received.

6. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to select the summary structure based on at least one of:

a data entropy;

a skew;

a memory budget; or

a data attribute use frequency.

7. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to dynamically adjust the selected summary structures for the aggregated data flows based on newly received network traffic data.

8. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to condense the summary based on a memory budget.

9. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to compute the data distribution per a set of data attributes.

10. The apparatus according to claim 1 , wherein the instructions are further executable by the processor to identify and condense outdated summaries.

11. A computer-implemented method comprising:

accessing, by a processor, network traffic data pertaining to data flows among nodes in a network;

identifying, by the processor, features of the data flows;

partitioning, by the processor, the network traffic data into a plurality of feature windows based on the identified features of the data flow;

for each of the plurality of windows, aggregating, by the processor, data flows between pairs of nodes;

computing, by the processor, a data distribution of each of the aggregated data flows per a set of data attributes;

selecting, by the processor, a summary structure for each of the aggregated data flows based on the computed data distributions of the aggregated data flows;

generating, by the processor, a summary of each of the aggregated data flows using the selected summary structures for the aggregated data flows

classifying, by the processor, queries applied to a summary of an aggregated data flow; and

selecting, by the processor, a new summary structure for the aggregated data flow based on the classified queries.

12. The method according to claim 11 , further comprising selecting the summary structure from a library of summary structures.

13. The method according to claim 11 , further comprising:

maintaining the selected summary structure for each of the aggregated data flows until new network traffic data is accessed.

14. The method according to claim 11 , further comprising:

selecting the summary structures for the aggregated data flows based on at least one of:

a data entropy;

a skew;

a memory budget; or

a data attribute use frequency.

15. The method according to claim 11 , further comprising:

dynamically adjusting the selected summary structures for the aggregated data flows based on newly received network traffic data.

16. A non-transitory computer readable medium on which is stored machine readable instructions that when executed by a processor, cause the processor to:

access network traffic data pertaining to data flows among nodes in a network;

partition the network traffic data into a plurality of windows;

for each of the plurality of windows, aggregate data flows between pairs of nodes;

compute a data distribution of each of the aggregated data flows;

determine a memory budget;

select a summary structure for each of the aggregated data flows based on the computed data distributions of the aggregated data flows and the determined memory budget;

generate a summary of each of the aggregated data flows using the selected summary structures for the aggregated data flows;

store the generated summaries;

classify queries applied to a summary of an aggregated data flow; and

select a new summary structure for the aggregated data flow based on the classified queries.

17. The non-transitory computer readable medium according to claim 16 , wherein the instructions are further to cause the processor to:

maintain the selected summary structure for each of the aggregated data flows until new network traffic data is accessed.

18. The non-transitory computer readable medium according to claim 16 , wherein the instructions are further to cause the processor to:

select the summary structures for the aggregated data flows based on at least one of:

a data entropy;

a skew;

the memory budget; or

a data attribute use frequency.