Managing runtime execution of applications on cloud computing systems

• US 8,719,804 B2
• Filed: 05/05/2010
• Issued: 05/06/2014
• Est. Priority Date: 05/05/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Jain, Navendu
• Primary Examiner(s): Dao, Thuy
• Assistant Examiner(s): Nguyen, Mongbao
• Application Number:
  US12/774,203
• Publication Number:
  US 20110276951A1
• Time in Patent Office: 1,462 Days
• Field of Search: None
• US Class Current: 717/140
• CPC Class Codes: G06F 11/3006 where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems multiprogramming arrangements G06F9/46; allocation of resources G06F9/50
• CPC Class Codes: G06F 11/301 where the computing system is a virtual computing platform, e.g. logically partitioned systems virtual machines G06F9/45533; logical partitioning of resources G06F9/5077
• CPC Class Codes: G06F 11/3082 the data filtering being achieved by aggregating or compressing the monitored data
• CPC Class Codes: G06F 11/3089 Monitoring arrangements determined by the means or processing involved in sensing the monitored data, e.g. interfaces, connectors, sensors, probes, agents software debugging using additional hardware using a specific debug interface G06F11/3656; performance evaluation by tracing or monitoring G06F11/3466
• CPC Class Codes: G06F 11/3476 Data logging G06F11/14, G06F11/2205 take precedence
• CPC Class Codes: G06F 11/3495 for systems
• CPC Class Codes: G06F 2201/865 Monitoring of software
• CPC Class Codes: H04L 43/0829 Packet loss
• CPC Class Codes: H04L 43/0894 Packet rate

Abstract

Instances of a same application execute on different respective hosts in a cloud computing environment. Instances of a monitor application are distributed to concurrently execute with each application instance on a host in the cloud environment, which provides user access to the application instances. The monitor application may be generated from a specification, which may define properties of the application/cloud to monitor and rules based on the properties. Each rule may have one or more conditions. Each monitor instance running on a host, monitors execution of the corresponding application instance on that host by obtaining from the host information regarding values of properties on the host per the application instance. Each monitor instance may evaluate the local host information or aggregate information collected from hosts running other instances of the monitor application, to repeatedly determine whether a rule condition has been violated. On violation, a user-specified handler is triggered.

Claims

1. A computer-implemented method of managing applications hosted in a cloud computing environment, where application instances, comprising instances of the same application, execute on different respective hosts in the cloud computing environment, the method comprising:

distributing, to each of the hosts, to concurrently execute with each application instance a corresponding monitor application instance, such that each application instance on a given host is paired with a monitor application instance on the given host in the cloud computing environment, the monitor application comprising an application that was generated from a specification of runtime properties and characteristics of the application and the cloud computing infrastructure, the specification defining properties of the application and the cloud computing infrastructure that are to be monitored and rules based on the properties and characteristics, each rule comprising one or more conditions, and wherein one of the properties comprise an aggregate property of the application;

executing each paired application instance and monitor instance on a different host in the cloud computing environment, wherein the cloud computing environment provides access to the application instances to users outside the cloud computing infrastructure;

each monitor application concurrently running on a host with an application instance monitors execution of the corresponding application instance, by obtaining from the host information regarding usage of resources on the host by the application instance, runtime properties, characteristics and performance metrics of the cloud computing infrastructure, and properties and performance metrics exported by that application instance and other application instances sharing common resources on the host;

each of the monitor applications using the host information to repeatedly determine whether a rule condition in the specification has been satisfied; and

collecting corresponding values of the aggregate properties from the monitor instances and determining whether a condition of a rule in the application specification is satisfied by an aggregation of the collected values.

2. A computer-implemented method according to claim 1, wherein the collecting is performed using a distributed overlay network formed by the monitor application instances, each monitor application instance comprising a node in the distributed overlay network.

3. A computer-implemented method according to claim 2, wherein one or more of the applications run in virtual machines on respective of the hosts.

4. A computer-implemented method according to claim 1, wherein a rule includes a frequency that controls how often each monitor application instance will determine whether the condition of the rule has been satisfied, and a property includes one or more of:

a frequency that controls how often each monitor application will obtain from its host a value of resource usage on the host, a target set of nodes or ports or protocols or IP addresses or combinations thereof, an aggregate over a set of output values from sensors, a number of file handles used, IO read/write rates, a page files, number of processes, number of threads, network loss rate, availability of an IP address, number of messages sent/received, average bandwidth used, or a runtime property and characteristic of the cloud computing infrastructure.

5. A computer-implemented method according to claim 1, wherein the collecting is performed using a centralized monitoring system where monitor application instances may provide their local data for aggregation and checking of specified properties.

6. A computer-implemented method according to claim 1, wherein a rule in the application specification comprises indicia of a violation handler, and each monitor instance invokes the violation handler on its host when the condition of the rule is satisfied on that host.

7. A computer-implemented method according to claim 1, wherein the application specification is written in a specification language, and generating of the monitor comprises passing the application specification to a compiler that is able to compile any arbitrary specification written in the specification language into a corresponding executable application.

8. One or more computer-readable storage memory devices storing information to enable host computers in a cloud computing network to perform a process, wherein the computer-readable storage media is not a signal, wherein host computers of the cloud computing network host arbitrary applications of various users by distributing the arbitrary applications among the host computers, the host computers provided with connectivity by a data network, the process comprising:

executing a plurality of application instances on respective host computers in the cloud computing network, each application instance comprising a copy of a same application program, where users access the application instances on the host computers via the cloud computing network, the cloud computing network providing access to the application instances by the users via the cloud computing network;

executing a plurality of monitor instances, one on each of the host computers hosting the application instances, respectively, wherein on any given of the host computers hosting a given monitor instance and given application instance, the monitor instance monitors effects of execution of the given application instance on the given host computer, each of the monitor instances comprising an instance of a same program compiled from a same specification that describes different types of runtime properties or characteristics of an element and/or aggregations thereof, the element comprising a component of the cloud computing network including the data network and/or the host computers and/or the application program; and

obtaining, by the application monitor instances, values of the properties and characteristics, and transmitting indicia of the values.

9. One or more computer-readable storage memory devices according to claim 8, wherein a plurality of monitor instances corresponding to all or the subset of the executing application instances obtain values of an aggregate property on the host computers where they are executing, and monitor instances in the plurality form a peer-to-peer network where each monitor instance is able to communicate with other monitor instances and using the peer-to-peer network to forward the values to a particular monitor instance.

10. One or more computer-readable storage memory devices according to claim 9, wherein the particular monitor instance receives the forwarded values and repeatedly computes an aggregate value of the forwarded values, and wherein the specification further comprises a rule comprising a condition that includes the aggregate property, and the particular monitor instance monitors the rule by repeatedly checking whether the condition is met based on the computed aggregate values.

11. One or more computer-readable storage memory devices according to claim 8, wherein information on the values is collected at a centralized monitoring system where monitor application instances may provide their local data for aggregation and checking of specified properties at global scale.

12. One or more computer-readable storage memory devices according to claim 8, wherein the specification also describes recovery and/or enforcement actions to be taken when respective rules are satisfied or violated by values of local properties or aggregate properties obtained by monitor instances according to the specification.

13. One or more computer-readable storage memory devices according to claim 12, wherein a monitor instance running on a host computer repeatedly obtains values of the local properties from the host computer or aggregate values of the local properties across multiple host computers, and when the values indicate that a rule is violated, the host computer performs a recovery and/or enforcement action corresponding to the violated rule.

14. A method performed in a cloud computing infrastructure that includes host computers that host different applications, the cloud computing infrastructure managing which host computers host instances of which of the applications, the cloud computing infrastructure providing users with uniform access to the applications, the method comprising:

receiving a monitor application, the monitor application having been compiled by a compiler compiling a specification, where the compiler implements a specification language that provides a sensor construct that includes a sensor type field and a frequency field, and a plurality of sensor types for the sensor construct, each sensor type corresponding to a generic network usage property or a generic computation usage property, the specification corresponding to an application and/or the cloud computing infrastructure, the specification comprising text conforming to the specification language, the specification comprising a particular sensor construct describing identifying a particular one of the sensor types or an aggregation, wherein when the compiler compiles the specification to generate the monitor application it builds the monitor application according to the specification such that the monitor application, when executed, will periodically query any host computer on which it is running for values corresponding to the sensor type identified in the specification; and

distributing instances of the application corresponding to the specification into the cloud computing infrastructure, the distributing comprising determining which host computers should host the application, and assuring that any host computer hosting the application is also provided with an instance of the monitor application, wherein the monitor application obtains, from whichever host computer it is running on, values of different properties including network resource usage or computation resource usage by the application on such host computer along-with runtime properties and characteristics of the cloud computing infrastructure.

15. A method according to claim 14, wherein the monitor application includes or has access to a peer-to-peer component on the host computer, wherein the specification further includes a second specific sensor construct comprising a second specific sensor type and an aggregate statistic for the second specific sensor type, the method further comprising the compiler responding to the presence of the aggregate statistic by building the monitor application such that instances of the monitor application will use the peer-to-peer components thereof to share and aggregate values for the second specific sensor type obtained from the respective host computers on which they are running.

16. A method according to claim 14, wherein the monitor applications transmit the obtained values to a given one of the monitor application instances, and the given one of the monitor application instances computes an aggregate statistic from the values, wherein the aggregate statistic is combined with another aggregate statistic to compute a final aggregate statistic.

17. A method according to claim 14, wherein the specification includes a rule, the monitor application is compiled to be able to test the rule, and the method further comprises determining, by the given one of the monitor applications, based on the aggregate statistic, whether the rule is violated.

18. A method according to claim 17, wherein the specification defines an action associated with the rule, the monitor application is compiled to be able to trigger the action, the method further comprising responding to a determination that the rule is violated by triggering, by the given one of the monitor application instances, the action.

19. A method according to claim 14, wherein the specification language further comprises a rule construct with a frequency element and violation handler element, and the specification comprises a specific rule construct that comprises a rule and references the specific set of sensor constructs, and the compiling further comprises building the monitor application according to the specific rule construct such that monitor application, when executed, will periodically determine whether the rule is satisfied or violated by the obtained sensor output values.