AWS vs. Azure: A Deep Dive into Model Training – Part 2

Having established how these platforms deal with mission setup and information entry, in Half 2, we’ll look at the compute assets and runtime environments that energy the mannequin coaching jobs.

Compute

Compute is the digital machine the place your mannequin and code run. Together with community and storage, it is likely one of the basic constructing blocks of cloud computing. Compute assets sometimes characterize the biggest value part of an ML mission, as coaching fashions—particularly giant AI fashions—requires lengthy coaching occasions and sometimes specialised compute cases (e.g., GPU cases) with increased prices. Subsequently, Azure ML designs a devoted AzureML Compute Operator function (see particulars in Part 1) for managing compute assets.

Azure and AWS supply varied occasion varieties that differ within the variety of CPUs/GPUs, reminiscence, disk area and sort, every designed for particular functions. Each platforms use a pay-as-you-go pricing mannequin, charging just for lively compute time.

Azure virtual machine series are named in alphabetic order; for example, D household VMs are designed for general-purpose workloads and meet the necessities for many growth and manufacturing environments. AWS compute instances are additionally grouped into households primarily based on their function; for example, the m5 household accommodates general-purpose cases for SageMaker ML growth. The desk under compares compute cases supplied by Azure and AWS primarily based on their function, hourly pricing and typical use circumstances. (Please notice that the pricing construction varies by area and plan, so I like to recommend trying out their official web sites.)

Now that we’ve in contrast compute pricing in AWS and Azure, let’s discover how the 2 platforms differ in integrating compute assets into ML programs.

Azure ML

Computes are persistent assets within the Azure ML Workspace, sometimes created as soon as by the AzureML Compute Operator and reused by the info science crew. Since compute assets are cost-intensive, this construction permits them to be centrally managed by a task with cloud infrastructure experience, whereas information scientists and engineers can give attention to growth work.

Azure gives a spectrum of compute goal choices designated for ML growth and deployment, relying on the size of the workload. A compute occasion is a single-node machine appropriate for interactive growth and testing within the Jupyter pocket book surroundings. A compute cluster is one other sort of compute goal that spins up multi-node cluster machines. It may be scaled for parallel processing primarily based on workload demand and helps auto-scaling by configuring the parameter min_instances and max_instances. Moreover, there are severless compute, Kubernetes clusters, and containers which are match for various functions. Here’s a helpful visible abstract that helps you make the choice primarily based in your use case.

To create an Azure ML managed compute goal we create an AmlCompute object utilizing the code under:

• sort: use"amlcompute" for compute cluster. Alternatively, use "computeinstance" for single-node interactive growth and “kubernetes" for AKS clusters.
• identify: specify the compute goal identify.
• measurement: specify the occasion measurement.
• min_instances and max_instances (elective): set the vary of cases allowed to run concurrently.
• idle_time_before_scale_down (elective): robotically shut down the compute cluster when idle to keep away from incurring pointless prices.

# Create a compute cluster
cpu_cluster = AmlCompute(
    identify="cpu-cluster",
    sort="amlcompute",
    measurement="Standard_DS3_v2",
    min_instances=0,
    max_instances=4,
    idle_time_before_scale_down=120
)

# Create or replace the compute
ml_client.compute.begin_create_or_update(cpu_cluster)

As soon as the compute useful resource is created, anybody within the shared Workspace can use it by merely referencing its identify in an ML job, making it simply accessible for crew collaboration.

# Use the continued compute "cpu-cluster" within the job
job = command(
    code='./src',
    command='python code.py',
    compute='cpu-cluster',
    display_name='train-custom-env',
    experiment_name='coaching'
)

AWS SageMaker AI

Compute assets are managed by a standalone AWS service – EC2 (Elastic Compute Cloud). When utilizing these compute assets in SageMaker, it require builders to explicitly configure the occasion sort for every job, then compute cases are created on-demand and terminated when the job finishes. This method provides builders extra flexibility over compute choice primarily based on job, however requires extra infrastructure information to pick and handle the suitable compute useful resource. For instance, obtainable occasion varieties differ by job sort. ml.t3.medium and ml.t3.giant are generally used for powering SageMaker notebooks in interactive growth environments, however they don’t seem to be obtainable for coaching jobs, which require extra highly effective occasion varieties from the m5, c5, p3, or g4dn households.

As proven within the code snippet under, AWS SageMaker specifies the compute occasion and the variety of cases operating concurrently as job parameters. A compute occasion with the ml.m5.xlarge sort is created throughout job execution and charged primarily based on the job runtime.

estimator = Estimator(
    image_uri=image_uri,
    function=function,  
    instance_type="ml.m5.xlarge", 
    instance_count=1
)

SageMaker jobs spin up on-demand cases by default. They’re charged by seconds and gives assured capability for operating time-sensitive jobs. For jobs that may tolerate interruptions and better latency, spot occasion is a extra cost-saving possibility that makes use of unused compute cases. The draw back is the extra ready interval when there are not any obtainable spot cases. We use the code snippet under to implement a spot occasion possibility for a coaching job.

• use_spot_instances: set as True to make use of spot cases, in any other case default to on-demand
• max_wait: the utmost period of time you’re keen to attend for obtainable spot cases (ready time is just not charged)
  max_run: the utmost quantity of coaching time allowed for the job
• checkpoint_s3_uri: the S3 bucket URI path to avoid wasting mannequin checkpoints, in order that coaching can safely restart after ready

estimator = Estimator(
    image_uri=image_uri,
    function=function,  
    instance_type="ml.m5.xlarge", 
    instance_count=1,
    use_spot_instances=True, 
    max_run=3600,
    max_wait=7200,  
    checkpoint_s3_uri=""  
)

What does this imply in follow?

• Azure ML: Azure’s persistent compute method permits centralized administration and sharing throughout a number of builders, permitting information scientists to give attention to mannequin growth fairly than infrastructure administration.
• AWS SageMaker AI: SageMaker requires builders to explicitly outline compute occasion sort for every job, offering extra flexibility but in addition demanding deeper infrastructure information of occasion varieties, prices and availability constraints.

Reference

Setting

Setting defines the place the code or job is run, together with software program, working system, program packages, docker picture and surroundings variables. Whereas compute is chargeable for the underlying infrastructure and {hardware} alternatives, surroundings setup is essential in guaranteeing constant and reproducible behaviors throughout growth and manufacturing surroundings, mitigating bundle conflicts and dependency points when executing the identical code in numerous runtime setup by totally different builders. Azure ML and SageMaker each help utilizing their curated environments and organising {custom} environments.

Azure ML

Just like Knowledge and Compute, Setting is taken into account a kind of useful resource and asset within the Azure ML Workspace. Azure ML gives a complete record of curated environments for in style python frameworks (e.g. PyTorch, Tensorflow, scikit-learn) designed for CPU or GPU/CUDA goal.

The code snippet under helps to retrieve the record of all curated environments in Azure ML. They typically observe a naming conference that features the framework identify, model, working system, Python model, and compute goal (CPU/GPU), e.g.AzureML-sklearn-1.0-ubuntu20.04-py38-cpu signifies scikit-learn model 1.0, operating on Ubuntu 20.04 with Python 3.8 for CPU compute.

envs = ml_client.environments.record()
for env in envs:
    print(env.identify)
    
    
# >>> Auzre ML Curated Environments
"""
AzureML-AI-Studio-Improvement
AzureML-ACPT-pytorch-1.13-py38-cuda11.7-gpu
AzureML-ACPT-pytorch-1.12-py38-cuda11.6-gpu
AzureML-ACPT-pytorch-1.12-py39-cuda11.6-gpu
AzureML-ACPT-pytorch-1.11-py38-cuda11.5-gpu
AzureML-ACPT-pytorch-1.11-py38-cuda11.3-gpu
AzureML-responsibleai-0.21-ubuntu20.04-py38-cpu
AzureML-responsibleai-0.20-ubuntu20.04-py38-cpu
AzureML-tensorflow-2.5-ubuntu20.04-py38-cuda11-gpu
AzureML-tensorflow-2.6-ubuntu20.04-py38-cuda11-gpu
AzureML-tensorflow-2.7-ubuntu20.04-py38-cuda11-gpu
AzureML-sklearn-1.0-ubuntu20.04-py38-cpu
AzureML-pytorch-1.10-ubuntu18.04-py38-cuda11-gpu
AzureML-pytorch-1.9-ubuntu18.04-py37-cuda11-gpu
AzureML-pytorch-1.8-ubuntu18.04-py37-cuda11-gpu
AzureML-sklearn-0.24-ubuntu18.04-py37-cpu
AzureML-lightgbm-3.2-ubuntu18.04-py37-cpu
AzureML-pytorch-1.7-ubuntu18.04-py37-cuda11-gpu
AzureML-tensorflow-2.4-ubuntu18.04-py37-cuda11-gpu
AzureML-Triton
AzureML-Designer-Rating
AzureML-VowpalWabbit-8.8.0
AzureML-PyTorch-1.3-CPU
"""

To run the coaching job in a curated surroundings, we create an surroundings object by referencing its identify and model, then passing it as a job parameter.

# Get an curated Setting
surroundings = ml_client.environments.get("AzureML-sklearn-1.0-ubuntu20.04-py38-cpu", model=44)

# Use the curated surroundings in Job
job = command(
    code=".",
    command="python practice.py",
    surroundings=surroundings,
    compute="cpu-cluster"
)

ml_client.jobs.create_or_update(job)

Alternatively, create a {custom} surroundings from a Docker picture registered in Docker Hob utilizing the code snippet under.

# Get an curated Setting
surroundings = ml_client.environments.get("AzureML-sklearn-1.0-ubuntu20.04-py38-cpu", model=44)

# Use the curated surroundings in Job
job = command(
    code=".",
    command="python practice.py",
    surroundings=surroundings,
    compute="cpu-cluster"
)

ml_client.jobs.create_or_update(job)

AWS SageMaker AI

SageMaker’s surroundings configuration is tightly coupled with job definitions, providing three ranges of customization to ascertain the OS, frameworks and packages required for job execution. These are Constructed-in Algorithm, Deliver Your Personal Script (Script mode) and Deliver Your Personal Container (BYOC), starting from the simplest but inflexible choice to essentially the most complicated but customizable possibility.

Constructed-in Algorithms

That is the choice with the least quantity of effort for builders to coach and deploy machine studying fashions at scale in AWS SageMaker and Azure at present doesn’t supply an equal built-in algorithm method utilizing Python SDK as of February 2026.

SageMaker encapsulates the machine studying algorithm, in addition to its python library and framework dependencies inside an estimator object. For instance, right here we instantiate a KMeans estimator by specifying the algorithm-specific hyperparameter ok and passing the coaching information to suit the mannequin. Then the coaching job will spin up a ml.m5.giant compute occasion and the educated mannequin will probably be saved within the output location.

Deliver Your Personal Script

The deliver your personal script method (often known as script mode or deliver your personal mannequin) permits builders to leverage SageMaker’s prebuilt containers for in style python frameworks for machine studying like scikit-learn, PyTorch and Tensorflow. It gives the flexibleness of customizing the coaching job by means of your personal script with out the necessity of managing the job execution surroundings, making it the preferred alternative when utilizing specialised algorithms not included in SageMaker’s built-in choices.

Within the instance under, we instantiate an estimator utilizing the scikit-learn framework by offering a {custom} coaching script train.py, the mannequin’s hyperparameters, together with the framework model and python model.

from sagemaker.sklearn import SKLearn

sk_estimator = SKLearn(
    entry_point="practice.py",
    function=function,
    instance_count=1,
    instance_type="ml.m5.giant",
    py_version="py3",
    framework_version="1.2-1",
    script_mode=True,
    hyperparameters={"estimators": 20},
)

# Prepare the estimator
sk_estimator.match({"practice": training_data})

Deliver Your Personal Container

That is the method with the very best degree of customization, which permits builders to deliver a {custom} surroundings utilizing a Docker picture. It fits eventualities that depend on unsupported python frameworks, specialised packages, or different programming languages (e.g. R, Java and many others). The workflow entails constructing a Docker picture that accommodates all required bundle dependencies and mannequin coaching scripts, then push it to Elastic Container Registry (ECR), which is AWS’s container registry service equal to Docker Hub.

Within the code under, we specify the {custom} docker picture URI as a parameter to create the estimator and match the estimator with coaching information.

from sagemaker.estimator import Estimator

image_uri = ":"

byoc_estimator = Estimator(
    image_uri=image_uri,
    function=function,
    instance_count=1,
    instance_type="ml.m5.giant",
    output_path="",
    sagemaker_session=sess,
)

byoc_estimator.match(training_data)

What does it imply in follow?

• Azure ML: Gives help for operating coaching jobs utilizing its intensive assortment of curated environments that cowl in style frameworks similar to PyTorch, TensorFlow, and scikit-learn, in addition to providing the potential to construct and configure {custom} environments from Docker photos for extra specialised use circumstances. Nevertheless, you will need to notice that Azure ML doesn’t at present supply the built-in algorithm method that encapsulates and packages in style machine studying algorithms immediately into the surroundings in the identical manner that SageMaker does.
• AWS SageMaker AI: SageMaker is understood for its three degree of customizations—Constructed-in Algorithm, Deliver Your Personal Script, Deliver Your Personal Container—which cowl a spectrum of builders necessities. Constructed-in Algorithm and Deliver Your Personal Script use AWS’s managed environments and combine tightly with ML algorithms or frameworks. They provide simplicity however are much less appropriate for extremely specialised mannequin coaching processes.

In Abstract

Based mostly on the comparisons of Compute and Setting above together with what we mentioned in AWS vs. Azure: A Deep Dive into Model Training — Part 1 (Venture Setup and Knowledge Storage), we might have realized the 2 platforms undertake totally different design ideas to construction their machine studying ecosystems.

Azure ML follows a extra modular structure the place Knowledge, Compute, and Setting are handled as impartial assets and belongings throughout the Azure ML Workspace. Since they are often configured and managed individually, this method is extra beginner-friendly, particularly for customers with out intensive cloud computing or permission administration information. For example, an information scientist can create a coaching job by attaching an present compute within the Workspace without having infrastructural experience to handle compute cases.

AWS SageMaker has a steeper studying curve, as a number of companies are tightly coupled and orchestrated collectively as a holistic system for ML job execution. Nevertheless, this job-centric method gives clear separation between mannequin coaching and mannequin deployment environments, in addition to the power for distributed coaching at scale. By giving builders extra infrastructure management, SageMaker is nicely suited to large-scale information science and AI groups with excessive MLOps maturity and the necessity of CI/CD pipelines.

Take-Dwelling Message

On this sequence, we evaluate the 2 hottest cloud platforms Azure and AWS for scalable mannequin coaching, breaking down the comparability into the next dimensions:

• Venture and Permission Administration
• Knowledge storage
• Compute
• Setting

In Part 1, we mentioned high-level mission setup and permission administration, then talked about storing and accessing the info required for mannequin coaching.