Amazon Technologies, Inc.

Douglas Stewart Laurence

James Christopher Sorenson

Munif M Farhan

Venkatraghavan Srinivasan

Darin Lee Frink

Anthony Nicholas Liguori

Andrew Bruce Dickinson

Neha Shetty

Kevin Christopher Miller

Andrew Gary Hourselt

Richard Alexander Sheehan

Fan Zhang

Marwan Salah EL-Din Oweis

Akshay Suhas Vaidya

Thomas Grant Slatton

Marc John Brooker

Pradeep Vincent

Bryan James Donlan

Alazel Acheson

Ryan F Watson

Seth William Markle

Matti Juhani Oikarinen

Matteo Frigo

Mikhail Danilov

Christopher Andrew Stephens

Sameer Choudhary

Fusheng Yuan

Richard Michael Sears

Ning Liao

Kaiwen Qu

Josef Schiefer

Ruiwen Zhao

Michael Banfield

Munif F Farhan

Eric Jason Brandwine

Nafea Bshara

Tate Andrew Certain

Christopher Richard Jacques De Kadt

Matthew Graham Baldwin

Petr Praus

Karishma Chawla

Karmveer Veer Singh

Steven Collison

Colm MacCárthaigh

John Pender

Zaher Dannawi

Andrew C Schleit

Milos Faro

Andrey Vasilyevich Talnikov

Yang Xu

Phil Simko

Daniel Leon Gregory Gardner

James Christopher Iii Sorenson

James Christopher Sorenson Iii

Iii James Christopher Sorenson

Systems and methods are described for rotating keys in a trust store to be used by a group of peer devices for secure communications between the peers in the group. In some examples, a service, such as an identify authority service, may make a determination that a set of peers that individually trust at least one public key from a group of public keys satisfies a set of conditions. As a result of the determination, the service may update the plurality of public keys by at least removing at least one public key from the group of public keys and indicate the updated plurality of public keys to at least one of the peers in the group. The service may remove the at least one public key from the group upon determining that less than a threshold number of peers in the group use the at least one public key.

Identity authority

Contents of client-initiated handshake messages of a security protocol are obtained at a handshake processing offloader configured for an application. The offloader uses a first security artifact (which is inaccessible from a front-end request processor of the application) and the contents of the handshake messages to generate a second security artifact. The second security artifact is transmitted to the front-end request processor, which uses it to perform cryptographic operations for client-server interactions of the application.

Security protocol handshake offloading

A storage service is configured to receive one or more instructions specifying transformations that are to be applied to data sets stored by the storage service when the data sets are made available outside of particular storage locations within the storage service. In response to triggering events that make the data sets available outside of the particular storage locations, the storage services causes the transformations to be performed on the data sets prior to the data sets being accessible at one or more destination locations outside of the particular storage locations where the data sets are stored. In some embodiments, the transformations are performed on hardware included in the storage service or are performed on external hardware at the direction of the storage service.

Storage service supporting data transformations

A distributed load balancer in which a router receives packets from at least one client and routes packet flows to multiple load balancer (LB) nodes according to a per-flow hashed multipath routing technique. For a given packet flow, the LB nodes randomly select a server node as a target for the packet flow from among multiple server nodes and send a connection request to the server node. A load balancer module on the server node makes the decision on whether to accept or reject the connection based on one or more metrics indicating a respective server's current load. If the module accepts the connection request, a connection is established between the server and the client. Otherwise, the load balancer nodes may select another server node and try again. The connections established between clients and servers pass through the load balancer nodes, but are not terminated at the load balancer nodes.

Distributed load balancer

Systems and methods are described for reducing latency to service requests to execute code on an on-demand code execution system by maintaining snapshots of virtual machine instances in a ready state to execute such code. A user may submit code to the on-demand code execution system, which code depends on other software, such as an operating system or runtime. The on-demand code execution system can generate a virtual machine instance provisioned with the other software, and initialize the instance into a state at which it is ready to execute the code. The on-demand code execution system can then generate a snapshot of the state of the instance, and halt the instance. When a request to execute the code is received, the snapshot can be used to quickly restore the instance. The code can then be executed within the instance, reducing the need to initialize the instance or maintain the instance in an executing state.

Reducing execution times in an on-demand network code execution system using saved machine states

In accordance with a designation of a private alias endpoint as a routing target for traffic directed to a service from within an isolated virtual network of a provider network, a tunneling intermediary receives a baseline packet generated at a compute instance. The baseline packet indicates a public IP (Internet Protocol) address of the service as the destination, and a private IP address of the compute instance as the source. In accordance with a tunneling protocol, the tunneling intermediary generates an encapsulation packet comprising at least a portion of the baseline packet and a header indicating the isolated virtual network. The encapsulation packet is transmitted to a node of the service.

Private alias endpoints for isolated virtual networks

A first instance is caused to execute software code to perform a first portion of a workflow in response to receipt of a workflow request, and performance of the first portion results in submission of an operation request to an entity. A resume workflow request is received from the entity, where the resume workflow request includes a handle to a snapshot that corresponds to a state of execution of the software code and a response to the operation request to the entity. Using the handle to the snapshot and the response to the operation request, a second instance is caused to execute the software code from the first state to perform a second portion of the workflow. A workflow result is received from an instance that executes a last portion of the workflow, and the workflow is provided result in response to the workflow request.

Techniques for performing continuation workflows

Systems and methods are provided for managing execution budgets for computing tasks in distributed computing systems. Execution budgets may include a budgeted number of operation retries that may be performed by the distributed computing system in response to failure of an operation. Execution budgets may also or alternatively include a budgeted amount of time in which the distributed computing system may perform the operations of a computing task. When a distributed computing system exhausts the execution budget allotted for a computing task, then further execution of the computing task or portions thereof may be terminated.

Distributed execution budget management system

A request to perform a workflow is received. A first instance is caused to be instantiated to perform a first portion of the workflow. First information and a handle associated with a second snapshot is received from the first instance. The first information is processed to produce a first result. A second instance is caused to be instantiated based on the handle to perform a second portion of the workflow. Second information is received from the second instance. The second information is processed to produce a second result, and an operation is performed dependent at least on the first result or the second result.

Chaining virtual machines

Technologies are provided for storing data by alternating the performance of data write operations using multiple clusters of storage devices. Data is written to internal buffers of storage devices in one cluster while data stored in buffers of storage devices in another cluster is transferred to the storage devices' permanent storages. When available buffer capacity in a cluster falls below a specified threshold, data write commands are no longer sent the cluster and the storage devices in the cluster transfer data stored in their buffers to their permanent storages. While the data is being transferred, data write commands are transmitted to other clusters. When the data transfer is complete, the storage devices in the cluster can be scheduled to receive data write commands again. A cluster can be selected for performing a given data write request by matching the attributes of the cluster to parameters of the data write request.

Cluster based hard drive SMR optimization

Technologies are provided for retrieving service level indicators (SLIs) for storage devices, prioritizing data access using the SLIs, and scheduling storage device maintenance based on the SLIs. A host computer can retrieve an SLI for a storage device and determine, using the SLI, whether the storage device is performing service and maintenance operations and/or whether the storage device is in need of performing service and maintenance operations. The host computer can retrieve SLIs for multiple storage devices storing data redundantly and prioritize using the storage devices to access the data based on the SLIs. If the host computer determines that a storage device is in need of maintenance, based on its SLI, the host computer can schedule the storage device for maintenance and divert data access requests to other storage devices until the maintenance is complete.

Host-accessible storage device service level indicators

Techniques described herein enhance the durability of cryptographically protected communications sessions. The negotiation of a cryptographically protected communications session results in the negotiation of a primary secret and a secondary secret. The primary secret and secondary secret are stored in separate locations, such as in two locations in RAM, one of which being used as a RAM disk. The primary secret is used to cryptographically protect the communications session. Following the detection of a change of state event, the cryptographically protected communications session switches to the secondary secret in place of the primary secret to cryptographically protect the communications session.

Session resumption

A distributed load balancer in which a router receives packets from at least one client and routes packet flows to multiple load balancer nodes. The router exposes a public IP address and the load balancer nodes all advertise the same public IP address to the router. The router may implement a per-flow hashed multipath routing technique, for example an equal-cost multipath (ECMP) routing technique, to distribute the flows across the load balancer nodes. Thus, the multiple load balancer nodes may service a single public endpoint. The load balancer nodes may advertise to the router according to the Border Gateway Protocol (BGP). Rather than advertising itself, however, a load balancer node may be advertised to the router by one or more neighbor load balancer nodes; the neighbor nodes may terminate the BGP sessions with the router in response to determining that the load balancer node has failed.

Health checking in a distributed load balancer

Systems and methods are provided for managing computing tasks in distributed computing systems using execution budgets. Execution budgets may include a budgeted number of operation retries that may be performed by a distributed computing system in response to failure of an operation, or a budgeted amount of time in which the distributed computing system may perform the operations of a computing task. The configuration of the distributed system to execute the computing task may be generated based on prior consumption of execution budgets by portions of the computing system.

Dynamic distributed execution budget management system

Growing community of inventors

Mercer Island, WA, United States of America

Douglas Stewart Laurence