Amazon Technologies, Inc.

Ankit Kamboj

Vinayak Thapliyal

George Steven McPherson

Aravind Ramarathinam

Santosh Chandrachood

Amit Purohit

Ramkumar Kamalapuram Sugavanam

Linchi Zhang

Min Zhou

Gayatri Ramesh Deo

Lukasz Misiuda

Peter Sirota

Abhishek Rajnikant Sinha

Xing Wu

Jian Fang

Sumit Kumar

Vageesh Kumar

Dag Stockstad

In response to receiving a parallel processing job from a customer, a system operated by a computing resource service provider allocates and configures a cluster of computer systems capable of executing the job. In an embodiment, each computer system is configured with a first network stack that allows access to resources of the computing resource service provider and a second network stack that allows access to resources of the customer. In an embodiment, the state of the cluster is monitored by the system via the first network stack. In an embodiment, the system deploys a set of tasks on the cluster for fulfilling the processing job. In an embodiment, the tasks have access to the second network stack so that each task can access resources of the customer.

Cluster management

A system operated by a computing resource service provider maintains a pool of computing resources for use in processing job requests submitted by customers. The system allocates computing resources to the pool in anticipation of future resource needs. In various embodiments, future resource needs can be estimated based on scheduled jobs, or historical job information. In an embodiment, the computing resources are virtual computer systems which may be arranged in a cluster. In response to receiving a parallel processing job from a customer, the system reserves computing resources from the pool for performing the job. In an embodiment, the reserved resources are configured with a network namespace that is able to access to a customer's resources.

Resource management

Portions of distributed data processing programs may be dynamically allocated computing resources. When execution of a distributed data processing program is requested, interdependent portions of the resource may be dynamically computing resources according to requirements determined for the portions. A resource manager may, in some embodiments, respond to individual allocation requests for the different portions received from a program manager for the distributed data processing program. The allocated resources may be obtained from pools of computing resources that can be utilized by different distributed data processing programs.

Dynamically allocating resources for interdependent portions of distributed data processing programs

An indication of an input data stream comprising data records, stored at a stream management service, that are to be batched for a computation at a batch-oriented data processing service is received. A set of data records of the input data stream are identified, based on respective sequence numbers associated with the records, for a particular iteration of the computation. Metadata associated with the particular iteration, comprising identification information associated with the set of records on which the computation is performed during the particular iteration, is saved in a repository.

Stream adapter for batch-oriented processing frameworks

A system adapted to process long-running processes is disclosed. A request to upload data is received at a server. The server divides the data into multiple parts and launches a separate process to upload each of the divided parts. The server records for each process the processing time or duration that the particular process used to upload its corresponding data item. The server maintains an average processing duration that is calculated from the processing durations of the completed processes. The server identifies that one process is continuing to run and compares a processing duration for the particular process to a threshold derived from the average processing duration. If the processing duration for the particular process exceeds the threshold, the server initiates a new process to upload the same data item. When one of either the new process or the still running process has completed processing, the server terminates the other process.

Processing of long running processes

Growing community of inventors

Seattle, WA, United States of America

Ankit Kamboj