Abstract: A debugger core captures one or more inputs to a distributed application during live execution of the distributed application in a production environment. The debugger core applies the one or more inputs to the distributed application during replay execution of the distributed application in a debugging environment and performs debugging of the distributed application in the debugging environment.

Abstract: Methods and apparatus for flexible-location reservations and pricing for network-accessible resources are disclosed. A system includes a plurality of resources of a provider network distributed across multiple locations, and a resource manager. The resource manager implements a programmatic interface to allow a client to specify a flexible location option for a resource reservation request, indicating that the resource manager is to select one or more locations at which to reserve resource capacity. When a reservation request with the flexible location option specified is received, the resource manager selects a particular location based at least in part on heuristics using resource utilization data. In response to a resource activation request for the reservation, the resource manager activates a resource at a launch location selected from the multiple locations.

Abstract: Methods and apparatus for a cloud resource marketplace for third-party capacity are disclosed. A system includes a plurality of resources of a primary network, and a resource manager. The resource manager implements a reservation interface allowing a client to select and reserve resources of the primary network. In response to a resell request from a third-party resource provider, requesting advertisement of the availability of a third-party resource via the reservation interface, the resource manager may determine whether to resell the third-party resource based on one or more acceptance criteria. In response to determining to resell the third-party resource, the resource manager may indicate, via the reservation interface, a pricing policy associated with the third-party resource, an interruptibility setting of the third-party resource, and a level of compatibility of the third-party resource with a control interface implemented to allow clients to perform operations on resources of the primary network.

Abstract: Methods and apparatus for supporting instance ordering selection policies (ISOPs) for network-accessible resources are disclosed. A system includes a plurality of resource instances of a provider network, and a resource manager. The resource manager assigns an ISOP to be used to respond to instance acquisition requests of a specified type from a particular client. In response to a request of the specified type from the client, the resource manager determines, based at least in part on the ISOP, an order in which a plurality of resource groups (such as a set of instances reservations of the client, or an instance marketplace accessible to multiple clients) is to be examined to obtain an instance for the client. Starting with a first resource group, the resource manager examines each resource group until an instance matching instance selection criteria specified in the request is found.

Abstract: Methods and apparatus for flexible-location reservations and pricing for network-accessible resources are disclosed. A system includes a plurality of resources of a provider network distributed across multiple locations, and a resource manager. The resource manager implements a programmatic interface to allow a client to specify a flexible location option for a resource reservation request, indicating that the resource manager is to select one or more locations at which to reserve resource capacity. When a reservation request with the flexible location option specified is received, the resource manager selects a particular location based at least in part on heuristics using resource utilization data. In response to a resource activation request for the reservation, the resource manager activates a resource at a launch location selected from the multiple locations.

Abstract: Methods and apparatus for flexible-location reservations and pricing for network-accessible resources are disclosed. A system includes a plurality of resources of a provider network distributed across multiple locations, and a resource manager. The resource manager implements a programmatic interface to allow a client to specify a flexible location option for a resource reservation request, indicating that the resource manager is to select one or more locations at which to reserve resource capacity. When a reservation request with the flexible location option specified is received, the resource manager selects a particular location based at least in part on heuristics using resource utilization data. In response to a resource activation request for the reservation, the resource manager activates a resource at a launch location selected from the multiple locations.

Abstract: A computer system that provides pooled computing resources to clients may receive requests for resources, each of which includes a bid amount. While the supply of resources exceeds the demand, the system may adaptively determine a theoretical bid threshold for the resource pool that represents the predicted value of the lowest bid amount for which corresponding requests will be granted when demand for the resources exceeds supply. The determination may be dependent on a distribution of the bid amounts included in the requests for resources. The system may adaptively determine the lowest bid amount for which corresponding ones of the requests will be granted during non-contention dependent on changes in demand for the resources, the total capacity of the resource pool, and the theoretical bid threshold value, and may grant requests that include higher bid amounts. The use of some resources may be terminated in response to changes in demand.

Abstract: Methods and apparatus for a job resource planner for cloud computing environments are disclosed. A system includes a plurality of resource instances of a provider network, and a resource planner. The planner receives a plan request from a client, comprising a job goal and an indication of a resource acquisition policy to be used to obtain resource instances for the job. The policy specifies one or more instance data sources. The planner generates a resource acquisition plan for the job, based at least in part on an analysis of pricing data obtained from a specified data source. The analysis comprises one or more computation steps indicated in the policy. The generated plan includes at least one recommended acquisition price for a resource instance.

Abstract: A computer system that provides pooled computing resources to clients may receive requests for resources, each of which includes a bid amount. While the supply of resources exceeds the demand, the system may adaptively determine a theoretical bid threshold for the resource pool that represents the predicted value of the lowest bid amount for which corresponding requests will be granted when demand for the resources exceeds supply. The determination may be dependent on a distribution of the bid amounts included in the requests for resources. The system may adaptively determine the lowest bid amount for which corresponding ones of the requests will be granted during non-contention dependent on changes in demand for the resources, the total capacity of the resource pool, and the theoretical bid threshold value, and may grant requests that include higher bid amounts. The use of some resources may be terminated in response to changes in demand.