Amazon Technologies, Inc.

Marc J Brooker

Eric Jason Brandwine

David Everard Brown

Christopher Richard Jacques De Kadt

Marvin Michael Theimer

Abhinav Agrawal

Kevin Ross O'Neill

Mark Joseph Cavage

Cornelle Christiaan Pretorius Janse Van Rensburg

Peter Nicholas Desantis

Benjamin Tobler

Adam David Gray

Matthew Scott Garman

Erik James Fuller

Implementation resources are operated in a manner furthering a particular purpose while excluding use of the implementation resources for other purposes. At least some of the implementation resources have capacity that is usable to implement multiple other resources. The capacity of the implementation resources is allocated in a manner that satisfies one or more conditions on the capacity of the implementation resources that is used. Generally, the capacity is allocated in a manner that reduces the likelihood that resources initiated close in time will fail together should underlying implementation resources fail. The implementation resources may be hardware devices that implement virtual computer systems.

Resource allocation to reduce correlated failures

A plurality of virtual computing resources is detected to have been provisioned. Credentials are distributed to the plurality of virtual computing resources. A credentials map that maps the credentials to the plurality of virtual computing resources is updated. The credentials for the plurality of virtual computing resources are activated to enable the plurality of virtual computing resources to use the credentials to authenticate to a second computer system that manages a resource service, with the credentials being inaccessible to resources of the resource service. A virtual computing resource of the plurality of virtual computing resources is detected to been deprovisioned, and the credentials for the virtual computing resource are deactivated.

Techniques for credential generation

Techniques, including systems and methods, for virtual resource cost tracking account for unused capacity of implementation resources that are dedicated to particular customers and, as a result, are unusable for maintaining virtual resources for other customers. Customers requesting dedicated use of implementation resources are charged in a manner that compensates the virtual resource provider for the lost ability to use unused capacity of implementation resources to serve other customers. Customer charges may be determined by a pricing function that is calculated such that, for a base of customers, expected revenue from the base of customers approximates a revenue goal. The revenue goal may be determined based on revenue that would have been expected had the unused capacity of the dedicated implementation resources been available for serving other customers.

Statistically cost-following accounting model

Systems and methods for managing credentials distribute the credentials to subsets of a set of collectively managed computing resources. The collectively managed computing resources may include one or more virtual machine instances. The credentials distributed to the computing resources may be used by the computing resources to perform one or more actions. Actions may include performing one or more functions in connection with configuration, management, and/or operation of the one or more resources, and/or access of other computing resources. The ability to use credentials may be changed based at least in part on the occurrence of one or more events.

Techniques for managing credentials in a distributed computing environment

A system for allocating constrained resources (e.g., downstream services, execution threads, database connections, input/output channels, computational resources, and/or memory) to requested services that are dependent on those resources may include multiple resource queues, each of which maintains a queue of requests for a respective constrained resource, and multiple service request queues, from which requests may be subsequently serviced. As each request reaches the head of a resource queue, it may receive a resource token for a respective constrained resource. Once the request has collected resource tokens for each of the constrained resources on which it depends, the request may be passed to a service request queue that maintains a queue of requests of a particular type. Requests in the multiple service request queues may be serviced on a round-robin or weighted round-robin basis. The number of tokens available for each constrained resource may be modified based on observed system performance.

System and method for allocating resources for heterogeneous service requests

A provisioning system may select an available implementation resource by analyzing user data to infer a role of a proposed computing resource and comparing the role to roles of other existing computer resources. The user data may include a user-specified identifier for the proposed computing resource, such as a label. The role may be used to rank available implementation resources using factors including distance to, roles of, and relationship with other user computing resources. A ranking of available implementation resources may be used to select an implementation resource for provisioning the proposed computing resource.

Inferring resource allocation decisions from descriptive information

Techniques, including systems and methods, for virtual resource cost tracking account for unused capacity of implementation resources that is dedicated to particular customers and, as a result, is unusable for maintaining virtual resources for other customers. Customers requesting dedicated use of implementation resources are charged in a manner that compensates the virtual resource provider for the lost ability to use unused capacity of implementation resources to serve other customers. Customer charges may be determined by a pricing function that is calculated such that, for a base of customers, expected revenue from the base of customers approximates a revenue goal. The revenue goal may be determined based on revenue that would have been expected had the unused capacity of the dedicated implementation resources been available for serving other customers.

Statistically cost-following accounting model for dedicated resources

Systems and methods for attesting to information about a computing resource involve electronically signed documents. For a computing resource, a document containing information about the resource is generated and electronically signed. The document may be provided to one or more entities as an attestation to at least some of the information contained in the document. Attestation to information in the document may be a prerequisite for performance of one or more actions that may be taken in connection with the computing resource.

Techniques for attesting to information

Users intending to launch instances or otherwise access virtual resources in a multi-tenant environment can specify a launch configuration. For each type of instance or each type of user, at least one launch configuration is created that includes parameters and values to be used in instantiating an instance of that type, the values being optimized for the current environment and type of instance. Launch configurations can be optimized for different types of users, such as to account for security credentials and access levels. Such an approach enables users to launch instances by contacting the resource provider directly without need for a proxy, which can function as a choke point under heavy load. The use of an appropriate launch configuration can be enforced for any type of user at any level, such as at the sub-net level, by modifying a request that does not specify an appropriate launch configuration.

Enforceable launch configurations

In certain embodiments, a computer-implemented method includes accessing time series data that includes a number of values. The values of the time series data are transformed into multiple coefficient groups. Each coefficient group includes one or more coefficients. For each of the plurality of coefficient groups, quantized coefficient data for the coefficient group is determined. The quantized coefficient data for the plurality of coefficient groups is ordered according to one or more ordering policies to generate prioritized transfer data for communication over a communication network. The ordering polices may be determined according to a type of the time series data.

Formatting data for efficient communication over a network

Embodiments of systems and methods are described for dynamically tagging metrics data by a provider of computing resources. In some implementations, a requesting user or application can request the provider of computing resources to tag the metrics based on configurations and/or settings specified by the requesting user or application. The tagged metrics data can then be processed by a processing user or application at a later time.

Growing community of inventors

Cape Town, South Africa

Marc J Brooker