Modeling a computing entity

1. A method of modeling a computing entity, comprising:

receiving information relating to transactions associated with the computing entity, wherein the received information forms a collection of information;

segmenting, by at least one processor, the collection into a plurality of segments;

identifying, by the at least one processor, at least one anomalous segment in the plurality of segments;

filtering out the identified at least one anomalous segment;

providing corresponding segment models representing the respective segments other than the filtered out at least one anomalous segment of the collection;

combining the segment models for building a model of the computing entity;

detecting that a change has been made to the computing entity in response to determining that combining the segment models results in an error greater than a predefined threshold;

outputting the model of the computing entity in response to determining that combining the segment models does not result in error greater than the predefined threshold; and

outputting distinct models in response to determining that combining the segment model results in the error greater than the predefined threshold, where the distinct models include a first model based on the segment models before a point corresponding to the change, and a second model based on the segment models after the point corresponding to the change.

2. The method of claim 1 , wherein the collection of information comprises a time interval of information, and wherein segmenting the collection into the plurality of segments comprises segmenting the time interval into a plurality of time segments.

3. The method of claim 1 , wherein segmenting the collection into the plurality of segments comprises finding a segmentation of the collection that optimizes a predefined function relating to utilization of a resource by the computing entity.

4. The method of claim 3 , wherein finding the segmentation is further based on a penalty function that assigns penalties according to lengths of the segments.

5. The method of claim 1 , wherein identifying the at least one anomalous segment is based on detecting that a parameter associated with the at least one anomalous segment exceeds a threshold, where the parameter represents one or more of: (1) idle activities of a system in which the computing entity runs, and (2) a length of the anomalous segment.

6. The method of claim 1 , further comprising producing a cleansed log of transaction data for the computing entity, wherein the cleansed log is based on removing data associated with any anomalous segments.

7. The method of claim 1 , wherein the receiving, segmenting, identifying, filtering, providing, combining, and detecting are performed as part of an online environment in which the model of the computing entity is continually updated as additional information relating to execution of the computing entity is received.

8. A computer comprising:

storage media to store a collection of information regarding transactions performed by a computing entity;

at least one processor to:

receive information relating to transactions associated with the computing entity, wherein the received information forms a collection of information;

segment the collection into a plurality of segments;

identify at least one anomalous segment in the plurality of segments;

provide corresponding segment models representing the respective segments other than the at least one anomalous segment of the collection;

combine the segment models for building a model of the computing entity;

detect that a change has been made to the computing entity in response to determining that combining the segment models results in an error greater than a predefined threshold;

output the model of the computing entity in response to determining that combining the segment models does not result in error greater than the predefined threshold; and

output distinct models in response to determining that combining the segment model results in error greater than the predefined threshold, where the distinct models include a first model based on the segment models before a point corresponding to the change, and a second model based on the segment models after the point corresponding to the change.

9. The computer of claim 8 , wherein the at least one processor is configured to further:

identify one or more transactions of the computing entity that relate to the change.

10. The computer of claim 8 , wherein the collection includes a time interval, and the plurality of segments include a plurality of time.

11. An article comprising at least one computer-readable storage medium containing instructions that upon execution cause a computer to:

divide data associated with transactions of a computing entity into a plurality of segments;

identify at least one of the segments as being anomalous;

provide segment models for corresponding ones of the plurality of segments other than the identified at least one anomalous segment;

perform reconciliation that includes combining the segment models to produce an overall model;

detect that a change has been made to the computing entity that prevents the overall model from being an accurate model of the computing entity in response to determining that combining the segment models results in an error greater than a predefined threshold;

output the overall model in response to determining that combining the segment models does not result in the error greater than the predefined threshold; and

output distinct models in response to determining that combining the segment model results in the error greater than the predefined threshold, where the distinct models include a first model based on the segment models before a time point corresponding to the change, and a second model based on the segment models after the time point corresponding to the change.