Using machine learning to dynamically determine a protocol for collecting system state information from enterprise devices
A method includes receiving data collected from a plurality of managed devices in a plurality of data collections. The data collections are performed using a plurality of collection protocols. A trigger that generated each of given ones of the data collections is determined. The method further includes identifying a collection protocol of the plurality of collection protocols used for each of the given ones of the data collections, and determining one or more attributes of a plurality of attributes of the plurality of managed devices that have been collected using given ones of the collection protocols. A mapping is generated between the triggers, the collection protocols and the attributes using one or more machine learning algorithms. The generated mapping is used to predict one or more collection protocols of the plurality of collection protocols to use to collect data from one or more of the managed devices.
1. An apparatus comprising:
at least one processing platform comprising a plurality of processing devices;
said at least one processing platform being configured:
to receive data collected from a plurality of managed devices in a plurality of data collections, wherein the plurality of data collections are performed using a plurality of collection protocols;
to determine, for given ones of the plurality of data collections, a trigger of a plurality of triggers that generated each of the given ones of the plurality of data collections;
to identify a collection protocol of the plurality of collection protocols used for each of the given ones of the plurality of data collections;
to determine one or more attributes of a plurality of attributes that have been collected using given ones of the plurality of collection protocols, wherein the plurality of attributes are of the plurality of managed devices;
to generate a mapping between the plurality of triggers, the plurality of collection protocols and the plurality of attributes using one or more machine learning algorithms;
to predict, using the generated mapping, one or more collection protocols of the plurality of collection protocols to use to collect data from one or more of the plurality of managed devices;
to receive at least one of an alert and a warning detected on the one or more of the plurality of managed devices; and
to identify one or more of the plurality of attributes corresponding to the at least one of the alert and the warning;
wherein the prediction of the one or more collection protocols to use to collect the data from the one or more of the plurality of managed devices is based on the identified one or more of the plurality of attributes corresponding to the at least one of the alert and the warning.
2. The apparatus of claim 1 wherein the plurality of collection protocols comprise one or more of Simple Network Management Protocol (SNMP), Representational State Transfer (REST) protocol, Secure Shell (SSH) protocol and port 443 protocol.
3. The apparatus of claim 1 wherein the plurality of triggers comprise one or more of a periodic collection, an alert-based collection and a user-initiated collection.
4. The apparatus of claim 1 wherein the plurality of attributes correspond to a plurality of components of the plurality of managed devices.
5. The apparatus of claim 1 wherein the mapping comprises a decision tree.
6. The apparatus of claim 5 wherein said at least one processing platform is further configured to determine a plurality of weights of given nodes of the decision tree.
7. The apparatus of claim 6 wherein the weights of the given nodes of the decision tree are based on at least one of a type of one or more of the plurality of triggers and a number of the plurality of attributes collected by given ones of the plurality of collection protocols.
8. The apparatus of claim 7 wherein said at least one processing platform is further configured to rank the predicted one or more collection protocols based on the weights of the given nodes.
9. The apparatus of claim 1 wherein said at least one processing platform is further configured to use map/reduce techniques to partition the received data collected from the plurality of managed devices into a plurality of fuzzy sets.
10. The apparatus of claim 1 wherein the data collected from the plurality of managed devices comprises system state information.
11. The apparatus of claim 1 wherein said at least one processing platform is further configured to rank the predicted one or more collection protocols based on a number of the identified one or more of the plurality of attributes able to be retrieved from the one or more of the plurality of managed devices using a given collection protocol of the predicted one or more collection protocols.
12. The apparatus of claim 1 wherein said at least one processing platform is further configured:
to receive at least one of error data and activity log data collected from the one or more of the plurality of managed devices; and
to apply the at least one of the error data and the activity log data to the one or more machine learning algorithms to generate the mapping.
13. The apparatus of claim 1 wherein said at least one processing platform is further configured:
to receive one or more technical support tickets corresponding to the one or more of the plurality of managed devices; and
to apply the one or more technical support tickets to the one or more machine learning algorithms to generate the mapping.
14. A method comprising:
receiving data collected from a plurality of managed devices in a plurality of data collections, wherein the plurality of data collections are performed using a plurality of collection protocols;
determining, for given ones of the plurality of data collections, a trigger of a plurality of triggers that generated each of the given ones of the plurality of data collections;
identifying a collection protocol of the plurality of collection protocols used for each of the given ones of the plurality of data collections;
determining one or more attributes of a plurality of attributes that have been collected using given ones of the plurality of collection protocols, wherein the plurality of attributes are of the plurality of managed devices;
generating a mapping between the plurality of triggers, the plurality of collection protocols and the plurality of attributes using one or more machine learning algorithms;
predicting, using the generated mapping, one or more collection protocols of the plurality of collection protocols to use to collect data from one or more of the plurality of managed devices;
wherein the mapping comprises a decision tree; and
determining a plurality of weights of given nodes of the decision tree, wherein the weights of the given nodes of the decision tree are based on at least one of a type of one or more of the plurality of triggers and a number of the plurality of attributes collected by given ones of the plurality of collection protocols;
wherein the method is performed by at least one processing platform comprising at least one processing device comprising a processor coupled to a memory.
15. The method of claim 14 further comprising ranking the predicted one or more collection protocols based on the weights of the given nodes.
16. The method of claim 14 further comprising using map/reduce techniques to partition the received data collected from the plurality of managed devices into a plurality of fuzzy sets.
17. A computer program product comprising a non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing platform causes said at least one processing platform:
to receive data collected from a plurality of managed devices in a plurality of data collections, wherein the plurality of data collections are performed using a plurality of collection protocols;
to determine, for given ones of the plurality of data collections, a trigger of a plurality of triggers that generated each of the given ones of the plurality of data collections;
to identify a collection protocol of the plurality of collection protocols used for each of the given ones of the plurality of data collections;
to determine one or more attributes of a plurality of attributes that have been collected using given ones of the plurality of collection protocols, wherein the plurality of attributes are of the plurality of managed devices;
to generate a mapping between the plurality of triggers, the plurality of collection protocols and the plurality of attributes using one or more machine learning algorithms;
to predict, using the generated mapping, one or more collection protocols of the plurality of collection protocols to use to collect data from one or more of the plurality of managed devices;
wherein the mapping comprises a decision tree; and
to determine a plurality of weights of given nodes of the decision tree, wherein the weights of the given nodes of the decision tree are based on at least one of a type of one or more of the plurality of triggers and a number of the plurality of attributes collected by given ones of the plurality of collection protocols.
18. The computer program product according to claim 17 wherein the program code further causes said at least one processing platform to rank the predicted one or more collection protocols based on the weights of the given nodes.
19. The computer program product according to claim 17 wherein the program code further causes said at least one processing platform to use map/reduce techniques to partition the received data collected from the plurality of managed devices into a plurality of fuzzy sets.
20. The computer program product according to claim 17 wherein the program code further causes said at least one processing platform:
to receive at least one of error data and activity log data collected from the one or more of the plurality of managed devices; and
to apply the at least one of the error data and the activity log data to the one or more machine learning algorithms to generate the mapping.