logo
Published on

Impressive AWS features (I wish Azure Had)

cloud computing
Authors

Recently, while preparing for my AWS MLS-CO1 exam, I discovered some features that impressed me a lot. As someone with an extensive Azure background, I was pleasantly surprised to find that there may be more compelling reasons to make me lose my loyalty to Azure👀. From intelligent access policy generation based on CloudTrail Activity Logs to advanced A2I workflows, Intelligent Blob Access Tiers, and more. Lets dive into some of these features and how they compare to Azure's current capabilities.

Auto-Generated IAM Policies Based on CloudTrail Logs

When it comes to managing cloud resources for modern organisations, security is crucial. This is where AWS's ability to generate Identity and Access Management (IAM) policies based on CloudTrail logs stands out for me.

AWS’s Approach:

AWS CloudTrail is a service that enables governance, compliance, operational auditing, and risk auditing of your AWS account. By integrating CloudTrail logs with IAM policy generation, AWS provides a more automated and data-driven approach to policy management. Essentially, this feature analyzes user activities and API usage to recommend appropriate IAM policies following the principle of least privilege as per the AWS IAM Best Practices. This makes process not just simpler but more secure. This feature can be particularly useful for large organisations where IAM policies can be complex and difficult to track and manage.

Reference for Further Reading:

Azure’s Current Capabilities Comparatively, Azure offers robust policy and role-based access control mechanisms. Azure Policy helps enforce organizational standards and assess compliance at scale. However, the policy generation process is largely manual and doesn't dynamically adapt based on activity logs as AWS’s solution does. Azure Activity Log provides insights into subscription-level events but doesn’t directly integrate with policy generation in the same way.

Potential Benefits for Azure Automating policy generation based on user activities and API usage, like AWS, could lead to more tailored and secure access controls. This would not only bolster security but also reduce the administrative burden and potential human errors in policy configurations.

Reference for Azure Users:

AWS S3 Intelligent Blob Access Tiers

Did you know that AWS has a cost-savings feature called Intelligent Blob Tiers? It's a highly advanced data-driven system that optimizes your data storage efficiency by automatically updating your blob data lifecycle configuration. You won't have to worry about manually moving data around as this feature automatically shifts your data to the most cost-effective access tier based on how frequently it's being accessed.

AWS’s Intelligent Tiering AWS S3 Intelligent-Tiering is a storage class that delivers automatic cost savings by moving data to the most economical tier, based on usage patterns. It's designed for data with unknown or changing access patterns, making it ideal for long-term storage without the need to classify data based on its usage. I honestly wonder, why AWS would even offer this to customers? Does it not reduce their revenue? Very interesting decision! However, be that as it may I think dynamic tiering is a great feature for customers and I hope to see more of this from other cloud providers going forward.

Reference for Further Reading:

Azure Blob Storage: Current Capabilities In contrast, Azure Blob Storage offers several storage tiers (Hot, Cool, and Archive), but the transition between these tiers is mostly manual or based on fixed policies. While Azure Blob Storage is highly effective and reliable, the lack of an automated tiering system like AWS’s can lead to less optimized cost and efficiency, especially for data with unpredictable access patterns.

Potential Benefits for Azure This feature would allow Azure users to save on costs without the complexity of manually shifting data across different storage tiers. Azure is arguably the best cloud provider for data-driven AI solutions owing to therir recent partnership with OpenAI. I therefore expect more data-driven features like this to be added to Azure in the near future(Thank you Satya!)

Reference for Azure Users:

ML-Driven Cost Anomaly Detection

A critical aspect of cloud service management is cost control and optimization. AWS has taken a significant leap in this area with its ML-driven Cost Anomaly Detection feature, a tool that combines the power of Machine Learning with detailed cloud usage insights to identify unusual spending patterns.

AWS’s Innovative Cost Management AWS Cost Anomaly Detection harnesses machine learning to automatically monitor and analyze AWS spending. This feature flags unusual patterns and potential issues, providing near realtime alerts before the cost gets out of control. It's a proactive approach to cost management, allowing organizations to quickly identify and address unexpected charges, thereby avoiding budget overruns and optimizing cloud spending.

Reference for Further Reading:

Azure's Current Cost Management Tools Azure also provides comprehensive cost management tools, including Azure Cost Management, Analysis and potential Budget overrun predictions.

Potential Benefits for Azure By leveraging machine learning, Azure could provide more nuanced and predictive insights into spending patterns, empowering users to manage their cloud expenses more efficiently. This would be especially beneficial for large-scale enterprises where cloud spending is substantial and complex.

Reference for Azure Users:

A Better, More Extensible Annotation Solution

One of the areas where AWS excels is in providing data labelling solutions. It offers a range of options, from manual to semi-automated and fully automated solutions, to accelerate data annotation for supervised training. Additionally, there is a seamless integration with external labelling workforces if your private labelling team is not sufficient.

AWS’s Advanced Labeling Solutions AWS offers robust labeling features, especially notable in services like AWS SageMaker. SageMaker Ground Truth helps users build highly accurate training datasets for machine learning quickly. It supports a wide range of labeling tasks, including image, text, and 3D point cloud labeling. This service is not just about labeling; it's about doing so efficiently and at scale, with features like automated data labeling powered by machine learning and easy integration with other AWS services. Furthermore AWS offeres a seamless intergration with freelance labelling teams though AWS Mechanical Turk.

Reference for Further Reading:

Azure’s Labeling Capabilities While Azure offers several tools for data classification and management, its capabilities, particularly in terms of labeling solutions for machine learning, are not as extensive as AWS's. Azure Machine Learning does provide data labeling services, but these are(in my opinion) generally more basic and less integrated with machine learning processes compared to AWS's offering.

Potential Benefits for Azure A more comprehensive labeling solution would streamline the process of preparing large datasets, potentially offering features like AWS's automated data labeling and extensive integration capabilities. This would not only save time but also improve the overall quality of machine learning models developed on Azure.

Reference for Azure Users:

Augmented AI Features

Augmented AI is a growing field in data-driven solutions, blending traditional AI capabilities with human intelligence to enhance Machine Learning models'' prediction accuracy. This allow AI Enginners or developers to easily verify Machine Learning predictions by building quality control workflows to allow humans to assist the models where the prediction confidence is low.

AWS’s Approach to Augmented AI AWS offers a suite of Augmented AI services, particularly within Amazon SageMaker. These services allow developers to build, train, and deploy reliable Machine Learning models by incorporating human judgment into the inference workflows. For instance, Amazon SageMaker A2I (Augmented AI) integrates human reviews into machine learning pipelines, ensuring that AI predictions meet the quality standards and are continually improved based on human feedback.

Reference for Further Reading:

Azure’s Current AI Offerings Azure provides a range of AI and machine learning services, including Azure Machine Learning and various cognitive services. However, the platform’s integration of augmented AI isn’t as pronounced as AWS's. While Azure does offer tools for building and training machine learning models, the emphasis on blending these capabilities with human judgment is less evident.

Potential Benefits for Azure By integrating human insight into AI workflows, Azure could help businesses achieve more accurate and reliable AI outcomes, especially in scenarios where human judgment is crucial such as medical diagnosis or credit-card fraud detection.

Reference for Azure Users:

Closing thoughts

As someone deeply involved in Azure's ecosystem, recognizing the strengths of a rival platform like AWS has been both humbling and enlightening. It highlights the importance of cross-platform learning and the continuous pursuit of improvement in technology. With its recent partnership with OpenAI, Azure’s future in Machine Learning and Data Management looks bright, and with potential adoption or adaptation of these AWS features, it could shine even brighter.

How do you think Azure could integrate or improve upon these ideas? Let me know in the comments below.

Discuss on Twitter • View on GitHub