Identifying representative entities in clusters for it management

1. An apparatus comprising:

a processor; and

a memory on which is stored instructions that when executed by the processor, cause the processor to:

extract, from network traffic data, a connectivity matrix that identifies connectivity data between entities;

group the entities into a plurality of clusters based on the extracted connectivity matrix;

for each cluster of the plurality of clusters, identify a representative entity to represent the entities in the cluster, including calculating a coefficient for each entity in the cluster based on an average distance between the entity and other entities in the cluster and an average distance between the entity and entities in a neighboring cluster, and selecting an entity having a highest coefficient to be the representative entity of the cluster; and

for each cluster of the plurality of clusters, identify a communication behavior of the representative entity to represent a group behavior of all of the entities in the cluster, wherein the group behavior of the entities in each cluster is to be used for information technology management.

2. The apparatus of claim 1 , wherein the instructions further cause the processor to:

track the plurality of clusters over successive time windows to identify any emerging clusters, defunct clusters, reemerging clusters, or continuing clusters; and

output the identified emerging clusters, defunct clusters, reemerging clusters, or continuing clusters.

3. The apparatus of claim 2 , wherein to track the plurality of clusters over successive time windows, the instructions further cause the processor to:

for each of the successive time windows, group the entities into another plurality of clusters; and

identify the emerging clusters, defunct clusters, reemerging clusters, or continuing clusters from a comparison of the other plurality of clusters with the plurality of clusters.

4. The apparatus of claim 1 , wherein the instructions further cause the processor to:

reduce a dimensionality of the connectivity matrix; and

group the entities into the plurality of clusters based on the connectivity matrix that has the reduced dimensionality.

5. The apparatus of claim 1 , wherein the instructions that cause the processor to group the entities into the plurality of clusters include instructions that cause the processor to:

group the entities having similar communication behaviors with respect to each other into the plurality of clusters.

6. The apparatus of claim 1 , wherein the instructions further cause the processor to:

for each cluster of the plurality of clusters, identify a predefined number of top remote entities with which the entities in the cluster communicate, wherein the identified predefined number of top remote entities corresponds to the representative entity of the cluster.

7. The apparatus of claim 1 , wherein the instructions thither cause the processor to:

for each cluster of the plurality of clusters, identify a predefined number of top volume entities with which the entities communicate, wherein the identified predefined number of volume entities corresponds to the identified representative entity of the cluster.

8. The apparatus of claim 1 , wherein the instructions further cause the processor to:

determine the group behavior of the entities in each cluster of the plurality of clusters; and

apply a policy to all of the entities in each cluster based on the determined group behavior.

9. A method comprising:

identifying, by a processor, connectivity data between pairs of entities in a network;

grouping, by the processor, the entities into clusters based on the identified connectivity data between the pairs of entities;

for each cluster of the clusters, identifying, by the processor, a representative entity to represent the entities in the cluster, including calculating a coefficient for each entity in the cluster based on an average distance between the entity and other entities in the cluster and an average distance between the entity and entities in a neighboring cluster, and selecting an entity having a highest coefficient to be the representative entity of the cluster; and

for each cluster in the clusters, identifying, by the processor, a communication behavior of the representative entity to represent a group behavior of all of the entities in the cluster, wherein the group behavior of the entities in each cluster is to be used for application of an information technology management policy on the network.

10. The method of claim 9 , further comprising:

identifying connectivity data between the pairs of entities during a later time window; and

grouping the entities into new clusters based on the identified connectivity data between the pairs of entities during the later time window.

11. The method of claim 10 , further comprising:

identifying the new clusters grouped during the later time window as being an emerging cluster, a reemerging cluster, or a continuing cluster;

comparing the new clusters with the clusters grouped during a time window prior to the later time window; and

identifying, based on the comparison, whether any of the clusters grouped during the time window prior to the later time window is a defunct cluster.

12. The method of claim 11 , further comprising:

outputting an identification of the defunct cluster.

13. The method of claim 9 , further comprising:

extracting a connectivity matrix including the connectivity data between the pairs of entities;

reducing a dimensionality in the connectivity matrix; and

identifying frequencies of connections between the pairs of entities over the network based on data included in the connectivity matrix with the reduced dimensionality.

14. The method of claim 9 , further comprising:

for each cluster of the clusters, identifying a predefined number of top volume entities with which the entities in the cluster communicate, wherein the identified predefined number of top volume entities in the cluster corresponds to the representative entity of the cluster.

15. The method of claim 9 , further comprising:

for each cluster of the clusters, identifying a predefined number of top remote entities with which the entities in the cluster communicate, wherein the identified predefined number of top remote entities corresponds to the representative entity of the cluster.

16. The method of claim 9 , further comprising:

determining the group behavior for each cluster of the clusters; and

applying the information technology management policy to all of the entities in the cluster based on the determined group behavior of the cluster.

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

identify connectivity data between pairs of entities over a network during a first time window;

group the entities into clusters based on the identified connectivity data between the pairs of entities during the first time window;

for each cluster of the clusters, identify a representative entity that is representative of the entities in the cluster, including calculating a coefficient for each entity in the cluster based on an average distance between the entity and other entities in the cluster and an average distance between the entity and entities in a neighboring cluster, and selecting an entity having a highest coefficient to be the representative entity of the cluster; and

for each cluster of the clusters, identify a communication behavior of the representative entity to represent a group behavior of all of the entities in the cluster, wherein the group behavior of the entities in each cluster is to be used for application of an information technology management policy on the network.

18. The non-transitory computer-readable medium of claim 17 , wherein the instructions further cause the processor to:

identify connectivity data between the pairs of entities during a second time window; and

group the entities into new clusters based on the identified connectivity data between the pairs of entities during the second time window.

19. The non-transitory computer-readable medium of claim 18 , wherein the instructions further cause the processor to:

compare the new clusters grouped during the second time window with the clusters grouped during the first time window;

identify the new clusters as being an emerging cluster, a reemerging cluster, or a continuing cluster;

identify, based on the new clusters, whether any of the clusters grouped during the first time window is a defunct cluster; and

output an identification of the defunct cluster.

20. The non-transitory computer-readable medium of claim 17 , wherein the instructions further cause the processor to:

determine the group behavior of the entities in each cluster in the clusters; and

apply the information technology management policy to all of the entities in each cluster based on the determined group behavior.