AIM-AHEAD Service Workbench (SWB) on Amazon Web Services (AWS)
What is Service workbench on amazon web services?
Service Workbench on AWS
Service Workbench on AWS promotes infrastructure equity by providing the same analytical tools and level of access to researchers and students. The only infrastructure required to fully leverage this resource is a simple laptop and an internet connection.
SWB provides researchers and students with a user-friendly environment to configure and deploy their own secure cloud-computing environment in a few clicks. SWB is the first Apache 2.0 open-source native cloud computing platform that provides a modular and scalable solution to the supply of computing environments for researchers and students. The platform supplies students and research teams with a simple web application, empowering them to easily deploy and access any cloud workspace from a custom catalog of pre-configured and extendible) environments ( R, Jupyter Notebooks, Python, etc…) leveraging all AWS advanced data analysis tools and native security controls.
what is AIM-AHEAD Service WorkBench?
AIM-AHEAD Service Workbench (SWB)
SWB is the first open-source native cloud computing platform that provides a modular and scalable solution to the supply of research computing environments. The platform supplies research teams with a simple web application, empowering them to easily deploy and access any cloud workspace from a custom catalog of pre-configured environments leveraging all AWS AI/ML and native security controls.
SWB is a flexible and scalable cloud solution that promotes equitable access to the computational resources needed for AI/ML. Researchers have access to the compute power needed in a few clicks, regardless of the technical underlying complexity of it.
To get started with an analysis, end-users are only required to connect to the web application and select their desired configuration. The research workspace will be deployed in two clicks, selecting first the type of workspace and second the most applicable configuration for their analysis in terms of instance type, memory, CPUs, and GPUs.
Researchers will have access to the computing power they need, regardless of the technical underlying complexity of it. Moreover, there is a growing open community supporting various SWB workspaces, which enables the deployment of any type of computing workspaces.
AIM-AHEAD SERVICE WORKBENCH
What are studies on aim-ahead swb
Studies
Studies are linked to Amazon S3 buckets and provide data access and storage to users.
There are two types of studies available: Open Data Studies and Organization Studies.
Open Data Studies are available to all users through the Registry of Open Data on AWS.
Organization Studies can be private (only a single user can access) or shared across projects with multiple collaborators. Depending on your project, the studies you have access to may be empty or they may contain project-specific data. In either case, you must link studies to your workspace upon workspace creation in order to access data and save your work.
Researcher Workspace on AIM-AHEAD SWB
What is a research workspace?
Research workspaces are the foundation of a user’s interaction with SWB. These instances can be customized to a user’s needs, and include the following options:
- AWS SageMaker instances that work with widely used Jupyter Notebook formats
- RStudio server instances accessible over SSL
- EC2 Linux instances with SSH access
- EC2 Windows instances with RDP access
Workspace Configurations on AIM-AHEAD SWB
What are Workspace Configurations?
You can use different workspace configurations to customize the computing power of your instance. The standard workspace configurations and their pricing are outlined below:
Sagemaker workspace configurations:
Small |
Medium |
Large |
ml.t3.medium |
ml.m5.4xlarge |
ml.g5.4xlarge |
2 CPU |
16 CPU |
1 GPU |
4 GB |
64 GB |
16 CPU |
$0.05 / hr |
$0.922 / hr |
24 GB |
$1.20 / day |
$22.13 /day |
$1.62 / hr |
$38.98 / day |
EC2 (Rstudio, linux, windows) workspace configurations:
Small |
Medium |
Large |
t3.medium |
m5.2xlarge |
g4ad.4xlarge |
2 CPU |
8 CPU |
1 GPU |
4 GB |
32 GB |
16 vCPU |
$0.0416 / hr |
$0.48/hr |
64 GB |
$0.9984 / day |
$11.52/day |
$0.867 / hr |
|
|
$20.80800 / day |
These workspace configurations are fully customizable to researcher needs; see the below links for the full list of Sagemaker and EC2 configurations available. If you require a different workspace configuration for your research, contact us.
- Amazon EC2 instance types: https://aws.amazon.com/ec2/instance-types/
- Amazon Sagemaker configurations: https://aws.amazon.com/sagemaker/pricing/
examples
Resources Available -Out-of-the-Box- in SWB
AWS SageMaker instances that work with widely used Jupyter Notebook formats. Moreover, SageMaker instances come with staple ML/DL libraries (e.g., TensorFlow, PyTorch, MxNet), allowing savvy users to get started right away. On the opposite, non AI/ML experts can discover all pre-configured computing environments featuring many tutorials and analysis examples using public data in the form of Jupyter Notebooks for anyone to start learning using those resources. This AI/ML service accelerates innovation with purpose-built tools for every step of ML development, including labeling, data preparation, feature engineering, statistical bias detection, auto-ML, training, tuning, hosting, explainability, monitoring, and workflows. SageMaker helps data scientists and developers to prepare, build, train, and deploy high-quality ML models quickly by bringing together a broad set of capabilities purpose-built for ML, and enables the Fellows to develop and serve their own models.