Amazon Technologies, Inc.

Andrew James Lusk

Todd Vaughn Jonker

Chris A Suver

Eric Jason Brandwine

Marc John Brooker

Eric Paul Wei

Juan-Pierre Longmore

Anuj Prateek

James Alfred Gordon Greenfield

Christopher Richard Jacques De Kadt

Mikhail Danilov

Jacob Adam Gabrielson

Christopher Allen Suver

Michael Groenewald

Robin Alan Golden

Marc Andrew Bowes

Neil Andrew Eriksson

Jacobus Johannes Van Der Merwe

An application programming interface gateway service generates an application programming interface that, in various examples, allows client applications to access database functionality without maintaining active database connections, managing database credentials, or providing SQL code. The application programming interface maintains state information between invocations that allows for improved database performance. The state information may include SQL statements and subroutines, compiled SQL code, database credentials, active database connections, and connection pools. When invoked by a client application, the application programming interface may select an active database connection from a connection pool based at least in part on the activity history of each connection in the connection pool so that the expected cache performance of the database may be improved. Access to the application programming interface may be controlled via fine-grained access controls independent of the credentials used to access the database.

Stateful database application programming interface

Disclosed are various embodiments for centralized and dynamically generated service configurations for data center and other region builds. A computing environment may be configured to receive requests to deploy a new computing stack or service from different geographical regions at a central service. For the individual ones of the requests, the computing environment may identify computing resources of the computing environment required to deploy the new computing stack or service, determine an order of creation of the computing resources, and create at least one virtual process that automatically allocates the computing resources for the new computing stack based on a predetermined order of creation.

Centralized and dynamically generated service configurations for data center and region builds

Disclosed are various embodiments for automating data center builds through the use of service proxies. A proxy is configured in a testing computing environment to receive service calls for a plurality of first services in the testing computing environment. The proxy forwards the service calls to at least one corresponding instance of the plurality of first services not in the testing computing environment. One or more second services in the testing computing environment that generate the service calls for the plurality of first services are then tested. The service calls are handled by the proxy during testing.

Service proxies for automating data center builds

Systems and methods are described for management of garbage collection processes in an on-demand code execution system. An on-demand code execution system may execute user-submitted code on virtual machine instances. As each virtual machine instance executes code, garbage (e.g., memory that is allocated and then not properly de-allocated) may accumulate and may reduce the computing resources that the virtual machine instance makes available for allocation to further code executions. A garbage collection schedule manager may be used to generate a system-wide garbage collection schedule and manage garbage collection processes in accordance with the schedule. The garbage collection schedule manager may schedule garbage collection processes at times when demand for computing resources is forecasted to be low, and may coordinate scheduling of garbage collection processes across multiple virtual machine instances and host computing devices to prevent garbage collection processes from impacting code execution performance.

Network-level garbage collection in an on-demand code execution system

Launching data stores when new computing infrastructure is deployed is described. A proxy data storage service may be provided that receives instances of an application programming interface (API) call to store data and determines one of a first data store or a second data store to store the data. For instance, in response to receipt of a first instance of the API call, the proxy data storage service may store data in a first data store. However, after a provisioning of the second data store, the proxy data storage service may store data in the second data store in response to receipt of a second instance of the API call. In an instance in which the data stored in the second data store also exists in the first data store, the data is deleted from the first data store until an eventual migration of data is achieved.

Vectored data store for seeding and building data stores in new regions

A provider network provides task capture and execution for tasks performed with respect to resources in the provider network. Task actions performed with respect to resources may be captured based on inputs to a provider network via a network-based service for the provider network in order to generate a recorded task. The recorded task may be stored for subsequent execution. In response to the detection of an execution event for the recorded task, the task actions may be performed as described in the recorded task. In some embodiments, task parameters and execution events may be defined for recorded tasks. Multiple recorded tasks may be linked together to be managed and executed as recorded task workflows. Recorded tasks may also be obtained that were recorded by other clients, or recorded for different provider networks.

Capture and execution of provider network tasks

Systems and methods are described for management of garbage collection processes in an on-demand code execution system. An on-demand code execution system may execute user-submitted code on virtual machine instances, which may be hosted on various host computing devices. As each virtual machine instance executes code, garbage (e.g., memory that is allocated and then not properly de-allocated) may accumulate and may reduce the computing resources that the instance makes available for allocation to further code executions. A garbage collection manager may be implemented on each host computing device that implements garbage collection by taking a snapshot of the virtual machine instance prior to the accumulation of garbage, and then restoring the snapshot rather than identifying and de-allocating garbage on an item-by-item basis. The virtual machine instance may also identify state information that should be preserved, and the garbage collection manager may preserve or avoid overwriting this state information.

Snapshot-based garbage collection in an on-demand code execution system

A service workflow agent associated with a particular service of a provider network identifies a repository in which workflow descriptors indicative of tasks to be performed by the service are stored. In response to determining that a modification has been applied at one such repository, the agent examines a particular workflow descriptor stored in the repository. The agent initiates an attempt to implement a first task indicated in the descriptor. Subsequently, the agent updates the descriptor in accordance with an optimistic concurrency control technique to indicate a status of the first task.

Decentralized management of multi-service workflows

An API gateway hosts one or more APIs for use by client applications. The API gateway includes facilities for maintaining state information on behalf of a client application, a backend service, or an API. The state information may include cached information retrieved from backend services as well as cached results returned from the API to the customer applications. Cached results generated by the API may be shared across a number of customers or across a set of APIs managed by a single API gateway. Cached information associated with a backend service may be shared across a number of APIs that access the backend service. In some examples, session information associated with secure connections between an API gateway service and a backend service provider is maintained by the API gateway service and shared among APIs hosted by the API gateway service.

Application programming interface state management

An API provides a frontend interface to one or more backend services. Access to an API is controlled by a set of frontend credentials, and access to the one or more backend services is controlled by a set of backend credentials. A credential-translation table maintained within the API links each backend credential to one or more frontend credentials. Links between frontend and backend credentials may be managed by an administrator of the API. The API uses the translation table to translate frontend credentials provided with an API call into backend credentials used to access backend services. The API provides users with the ability to update the backend credentials in the credential-translation table based at least in part on the frontend credentials provided by the user. The API may limit the ability to extract backend credentials to administrative users.

Credential translation

A method and a system are provided for encoding and processing digital information. The digital information is encoded according to binary encoding formats corresponding to primitive data types. The primitive data types comprise scalar data types including Boolean, integer, float, decimal, time stamp, string, symbol, binary large object, and character large object data types. The primitive data types also comprise composite data types including structure, list, and S-expression data types. The binary-encoded digital information is stored in a message with a predetermined format for transmission. No metadata is included in the message.

Data format for processing information

A method and a system are provided for encoding and processing digital information. The digital information is encoded according to binary encoding formats corresponding to primitive data types. The primitive data types comprise scalar data types including Boolean, integer, float, decimal, time stamp, string, symbol, binary large object, and character large object data types. The primitive data types also comprise composite data types including structure, list and S-expression data types. The binary-encoded digital information is stored in a message with a predetermined format for transmission. No metadata is included in the message.

Growing community of inventors

Seattle, WA, United States of America

Andrew James Lusk