Method and system for coordinated service placement in multiple clouds
A service partitioned into service components requiring different resources is assigned to at least one cloud of a set of clouds. With interdependent service components, a service-placement engine is configured to coordinate assignments of individual service components. The service-placement engine receives from a client a definition of each service component and indications of components interdependence. Each cloud that satisfies compliance requirements, capability requirements, and resource requirements of a service component is considered eligible to host the service component. Selection of a specific eligible cloud is based on an overall cloud merit which depends on the service-component definition and the location of the client.
1. A method of assigning a multitask service to at least one cloud of a plurality of clouds, the method comprising:
employing a hardware processor to execute instructions for:
receiving from a client a set of service tasks, a definition of each service task, and indications of interdependence of the set of service tasks;
determining a dependency count of each task of the set of service tasks;
designating free tasks from the set of service tasks;
identifying a set of clouds from the plurality of clouds, wherein each of the identified clouds satisfies specified compliance requirements, capability requirements, and resource-availability requirements at a predetermined time for at least one of the designated free tasks;
determining an overall cloud merit vector for each of the free tasks and for the client;
sorting the identified set of clouds in descending order according to said overall cloud merit;
sequentially communicating with the identified set of clouds to request assignment of each of the free tasks, starting with the cloud of highest cloud merit;
assigning free tasks to respective clouds from the identified set of clouds;
receiving from the cloud confirmation of assignment of at least one of the service tasks;
determining that at least one of the assigned service tasks has a positive dependency count:
identifying succeeding tasks of at least one of the assigned tasks;
reducing the dependency count of each succeeding task by 1;
coordinating the assigned service tasks to respective clouds to observe the interdependence of the tasks, and returning to the designating a free task from the set of service tasks step.
2. The method of claim 1 further comprising:
acquiring from each cloud of the plurality of clouds respective cloud information;
determining for said each task a task-specific cloud merit of said each cloud according to said respective cloud information and said definition of each service task;
determining a proximity merit of said each cloud according to a known location of said client, and
determining said overall cloud merit vector according to said task-specific cloud merit and said proximity merit.
3. The method of claim 1 wherein said definition of each service task comprises: metadata; software instructions; and input data.
4. The method of claim 2 wherein said respective cloud information comprises at least one of the following:
a compliance vector indicating compliance with individual service standards of a predefined list of standards;
a capability vector indicating support of individual features of a predefined list of features;
a resource-availability vector indicating projected availability of resources; and
characterization data relevant to a predefined set of characteristics.
5. The method of claim 1 , wherein the set of service tasks comprises independent tasks.
6. The method of claim 1 , wherein the set of service tasks comprises interdependent tasks.
7. The method of claim 1 , wherein the set of service tasks comprises interdependent tasks, each of the interdependent tasks being subject to respective predetermined temporal constraints.
8. A system of assigning a multitask service to at least one cloud of a plurality of clouds, comprising:
a memory device having computer readable instructions stored thereon, for execution by a processor, causing the processor to:
receive from a client a set of service tasks, a definition of each service task, and indications of interdependence of the set of service tasks;
determine a dependency count of each task of the set of service tasks;
designating free tasks from the set of service tasks;
identify a set of clouds from the plurality of clouds; wherein each of the identified clouds satisfies specified compliance requirements, capability requirements, and resource-availability requirements at a predetermined time for at least one of the designated free tasks;
determine an overall cloud merit vector for each of the free tasks and for the client;
sort the identified set of clouds in descending order according to said overall cloud merit;
sequentially communicate with the identified set of clouds to request assignment of each of the free tasks, starting with the cloud of highest cloud merit;
assign free tasks to respective clouds from the identified set of clouds;
receive from the cloud confirmation of assignment of at least one of the service tasks;
determine that at least one of the assigned service tasks has a positive dependency count:
identify succeeding tasks of at least one of the assigned tasks;
reduce the dependency count of each succeeding task by 1;
coordinate the assigned service tasks to respective clouds to observe the interdependence of the tasks, and to return to the designating a free task from the set of service tasks step.
9. The system of claim 8 wherein the computer readable instructions further cause the processor to:
acquire from each cloud of the plurality of clouds respective cloud information;
determine for said each task a task-specific cloud merit of said each cloud according to said respective cloud information and said definition of each service task;
determine a proximity merit of said each cloud according to a known location of said client, and
determine said overall cloud merit vector according to said task-specific cloud merit and said proximity merit.
10. The system of claim 8 wherein said definition of each service task comprises: metadata; software instructions; and input data.
11. The system of claim 9 wherein said respective cloud information comprises at one of the following:
a compliance vector indicating compliance with individual service standards of a predefined list of standards;
a capability vector indicating support of individual features of a predefined list of features; a resource-availability vector indicating projected availability of resources; and
characterization data relevant to a predefined set of characteristics.