Aws::AcfInterface -- interface to Amazon CloudFront, a content distribution service
Aws::ElbInterface -- interface to Amazon Load Balancing service
Aws::MonInterface -- interface to Amazon CloudWatch monitoring service
Aws::Iam -- for AWS Identity and Access Management
To use a single piece intead of loading all of then, you can require it explicitly for example: require 'aws/sqs'.
Full programmmatic access to EC2, EBS, S3, SQS, SDB, ELB, and CloudFront.
Complete error handling: all operations check for errors and report complete error information by raising an AwsError.
Persistent HTTP connections with robust network-level retry layer using RightHttpConnection). This includes socket timeouts and retries.
Robust HTTP-level retry layer. Certain (user-adjustable) HTTP errors returned by Amazon's services are classified as temporary errors. These errors are automaticallly retried using exponentially increasing intervals. The number of retries is user-configurable.
Fast REXML-based parsing of responses (as fast as a pure Ruby solution allows).
Uses libxml (if available) for faster response parsing.
Support for large S3 list operations. Buckets and key subfolders containing many (> 1000) keys are listed in entirety. Operations based on list (like bucket clear) work on arbitrary numbers of keys.
Support for streaming GETs from S3, and streaming PUTs to S3 if the data source is a file.
Support for single-threaded usage, multithreaded usage, as well as usage with multiple AWS accounts.
Support for both first- and second-generation SQS (API versions 2007-05-01 and 2008-01-01). These versions of SQS are not compatible.
Support for signature versions 0, 1 and 2 on all services.
Test suite (requires AWS account to do "live" testing).
All AWS interfaces offer three threading options:
Use a single persistent HTTP connection per process. :single
Use a persistent HTTP connection per Ruby thread. :per_thread
Open a new connection for each request. :per_request
Either way, it doesn't matter how many (for example) Aws::S3 objects you create, they all use the same per-program or per-thread connection. The purpose of sharing the connection is to keep a single persistent HTTP connection open to avoid paying connection overhead on every request. However, if you have multiple concurrent threads, you may want or need an HTTP connection per thread to enable concurrent requests to AWS. The way this plays out in practice is:
If you have a non-multithreaded Ruby program, use the non-multithreaded setting.
If you have a multi-threaded Ruby program, use the multithreaded setting to enable concurrent requests to S3 (or SQS, or SDB, or EC2).
For running under Mongrel/Rails, use the non-multithreaded setting even though mongrel is multithreaded. This is because only one Rails handler is invoked at time (i.e. it acts like a single-threaded program)
Note that due to limitations in the I/O of the Ruby interpreter you may not get the degree of parallelism you may expect with the multi-threaded setting.
By default, EC2/S3/SQS/SDB/ACF interface instances are created in per_request mode. Set params[:connection_mode] to :per_thread in the initialization arguments to use multithreaded mode.