278daaeb7f
* Update to latest CLI/REST API * fix incorrect httpFiles text * Fix help text and incorrect change of mgmt_base_url to mgmt_batch_url * Update test recordings * removing trailing '/' from endpoint * Re record tests * Update os_flavor bug * Add blob source back as a map to http_url and update comments * remove excess url template option * Update to new python model scheme * pylint stuff * Fixup copy paste error for ParamtricSweepFactory * add license headers * Add template not implemented on exceed maximum api version * Add template api check to one more location and add tests * Update samples to 2018 |
||
---|---|---|
.. | ||
blender | ||
blender-appTemplate | ||
ffmpeg | ||
hello-world | ||
mpi | ||
ocr | ||
sdk | ||
readme.md |
readme.md
Azure Batch Samples
Here are a collection of samples to demonstrate the new features in this Batch preview CLI.
Hello World Samples
These samples use the new features in very simple ways to make it easier to see how these features might fit into your workflow.
Create pool and job
Create a pool and then run a job with a single task. Both the pool and the job are defined using templates with hard coded values.
Features used:
Create pool and job with templates
Create a pool and then run a job with a single task. Both the pool and the job are defined using a parameterized templates. Parameter values used to fill out the templates are stored in separate files that are easy to modify as required.
Features used:
- Pool and job templates with parameterization
- Parametric sweep task factory
- Task per file task factory
Task per file
Run a specific piece of processing independently across a set of files that are uploaded into storage. The job is specified as a template accepting parameters.
Features used:
- Input data upload to Batch linked storage accounts
- Pool and job templates with parameterization
- Task per file task factory
- Automatic persistence of task output files to Azure Storage
Task per file with application template
Run a specific piece of processing independently across a set of files that are uploaded into storage. The actual processing involved is split out into a separate application template. The job itself references the template while specifying parameters, pool information and other management details. Application templates are intended to be flexible and reusable across a number of jobs.
Features used:
- Split job configuration and management with reusable application templates
- Task per file task factory
- Automatic persistence of task output files to Azure Storage
More Complex Samples
These samples show how to use the new features with real world applications.
FFmpeg
FFmpeg is an open-source command line tool for processing multimedia files. This is a sample demonstrating audio compression with Azure Batch on a large number of numerically-named files using a parametric sweep.
Features used:
- Job template with parameterization
- Automatic persistence of task output files to Azure Storage
- Easy software installation via package managers
- Parametric sweep task factory
- Task per file task factory
OCR
OCR (Optical Character Recognition) is the process of extracting text from PDF images. This sample demonstrates the batch processing of PDF files.
Features used:
- Pool and job templates with parameterization
- Parametric sweep task factory
- Automatic persistence of task output files to Azure Storage
- Easy software installation via package managers
MPI
This sample demonstrates the batch run a MPI task with MultiInstanceSettings feature.
Blender
Blender is an open-source 3D content creation suite. This sample demonstrates distributed rendering on Azure Batch.
Features used:
- Job template with parameterization
- Parametric sweep task factory
- Automatic persistence of task output files to Azure Storage
- Easy software installation via package managers
Blender (Application Template)
A variation of the Blender sample that uses an application template to separate job definiton and management.
Features used: