Analyzing servers based on data streams generated by instrumented software executing on the servers
An instrumentation analysis system processes data streams received from servers executing instrumented software. The system determines a set of servers that satisfy a given criteria, for example, a set of servers with high resource utilization. The set of servers may be determined by the system based on triggers or specified by a user. The system analyzes properties of servers to determine a property that characterizes the set of servers. The property characterizing the servers is provided to users via a user interface or alerts for further analysis, for example, to analyze the cause of high resource utilization.
1. A method for analyzing servers executing instrumented software, the method comprising:
identifying a filtered set of servers from a plurality of servers, the filtered set of servers satisfying a filtering criterion;
selecting a plurality of properties of the filtered set of servers, each property from the plurality of properties comprising at least a name value pair having at least a dimension of servers and a value of the dimension;
for each property from the selected plurality of properties, determining a score based on a comparison of a likelihood of a server from the filtered set of servers having the property with a likelihood of a server outside the filtered set of servers having the property;
identifying a characteristic property of the filtered set of servers based on a ranking of the scores, such that servers from the filtered set of servers have a higher likelihood of having the characteristic property than servers outside the filtered set of servers; and
for each property representing a particular value for a particular dimension, ranking properties of servers, the ranking comprising:
determining a first value of a probability mass function for the particular dimension for servers in the filtered set; and
determining a second value of the probability mass function for the particular dimension for a set of servers comprising one or more servers outside the filtered set.
2. The method of claim 1 , further comprising:
storing attributes of each of the plurality of servers, each attribute associated with a dimension describing the servers, wherein each attribute value for a server is determined based on one of: data received as a data stream from the server or metadata describing the server specified independent of the data streams; and
receiving a filtering criterion based on an attribute representing resource utilization of servers, the filtering criterion determining whether the attribute representing resource utilization of servers has a value within a specified set of values.
3. The method of claim 2 , wherein the selected plurality of properties represents values of attributes based on metadata describing servers, the metadata specified independent of the data streams.
4. The method of claim 2 , wherein the resource utilization comprises at least one of CPU utilization, memory utilization, disk utilization, or network utilization.
5. The method of claim 1 , wherein the ranking further comprises:
determining a score for the property as a function of the first value of the probability mass function and the second value of the probability mass function.
6. The method of claim 5 , wherein the score for the property is a difference of the first value of the probability mass function and the second value of the probability mass function.
7. The method of claim 5 , wherein the score for the property is a ratio of the first value of the probability mass function and the second value of the probability mass function.
8. The method of claim 1 , further comprising:
generating an alert describing the characteristic property; and
sending the generated alert to a user account associated with at least one server of the plurality of servers.
9. The method of claim 1 , further comprising:
ranking the selected plurality of properties based on the scores.
10. A system for analyzing servers executing instrumented software, the system comprising:
at least one memory having instructions stored thereon; and
at least one processor configured to execute the instructions, wherein the at least one processor is configured to:
identify a filtered set of servers from a plurality of servers, the filtered set of servers satisfying a filtering criterion;
select a plurality of properties of the filtered set of servers, each property from the plurality of properties comprising at least a name value pair having at least a dimension of servers and a value of the dimension;
for each property from the selected plurality of properties, determine a score based on a comparison of a likelihood of a server from the filtered set of servers having the property with a likelihood of a server outside the filtered set of servers having the property;
identify a characteristic property of the filtered set of servers based on a ranking of the scores, such that servers from the filtered set of servers have a higher likelihood of having the characteristic property than servers outside the filtered set of servers; and
for each property representing a particular value for a particular dimension, ranking properties of servers, the ranking comprising:
determining a first value of a probability mass function for the particular dimension for servers in the filtered set; and
determining a second value of the probability mass function for the particular dimension for a set of servers comprising one or more servers outside the filtered set.
11. The system of claim 10 , further configured to:
store attributes of each of the plurality of servers, each attribute associated with a dimension describing the servers, wherein each attribute value for a server is determined based on one of: data received as a data stream from the server or metadata describing the server specified independent of the data streams; and
receive a filtering criterion based on an attribute representing resource utilization of servers, the filtering criterion determining whether the attribute representing resource utilization of servers has a value within a specified set of values.
12. The system of claim 11 , wherein the selected plurality of properties represents values of attributes based on metadata describing servers, the metadata specified independent of the data streams.
13. The system of claim 11 , wherein the resource utilization represents one of: CPU utilization, memory utilization, disk utilization, or network utilization.
14. The system of claim 10 , wherein the ranking further comprises:
determining a score for the property as a function of the first value of the probability mass function and the second value of the probability mass function.
15. The system of claim 14 , wherein the score for the property is a difference of the first value of the probability mass function and the second value of the probability mass function.
16. The system of claim 14 , wherein the score for the property is a ratio of the first value of the probability mass function and the second value of the probability mass function.
17. The system of claim 10 , further configured to:
generate an alert describing the characteristic property; and
send the generated alert to a user account associated with at least one server of the plurality of servers.
18. A non-transitory computer-readable storage medium comprising instructions stored thereon, which when executed by one or more processors, cause the one or more processors to perform operations for analyzing servers executing instrumented software, comprising:
identifying a filtered set of servers from a plurality of servers, the filtered set of servers satisfying a filtering criterion;
selecting a plurality of properties of the filtered set of servers, each property from the plurality of properties comprising at least a name value pair having at least a dimension of servers and a value of the dimension;
for each property from the selected plurality of properties, determining a score based on a comparison of a likelihood of a server from the filtered set of servers having the property with a likelihood of a server outside the filtered set of servers having the property;
identifying a characteristic property of the filtered set of servers based on a ranking of the scores, such that servers from the filtered set of servers have a higher likelihood of having the characteristic property than servers outside the filtered set of servers; and
for each property representing a particular value for a particular dimension, ranking properties of servers, the ranking comprising:
determining a first value of a probability mass function for the particular dimension for servers in the filtered set; and
determining a second value of the probability mass function for the particular dimension for a set of servers comprising one or more servers outside the filtered set.
19. The non-transitory computer-readable storage medium of claim 18 , further configured for:
generating an alert describing the characteristic property; and
sending the generated alert to a user account associated with at least one server of the plurality of servers.
20. The non-transitory computer-readable storage medium of claim 18 , further configured for:
storing attributes of each of the plurality of servers, each attribute associated with a dimension describing the servers, wherein each attribute value for a server is determined based on one of: data received as a data stream from the server or metadata describing the server specified independent of the data streams; and
receiving a filtering criterion based on an attribute representing resource utilization of servers.