Do you know what’s cooking in AWS?
The only constant thing in our lives is change. And Amazon is living proof of that statement.
In this post, I try to summarize all changes introduced in AWS over the last six months in a visual way so that you can quickly get an idea of what is happening in AWS and where they put their attention.
What is this about?
I will review the data from the “AWS What’s New” RSS feed, which I carefully collected for several months and prepared for analysis.
The data processing is built on AWS and composed of the following components:
- AWS Lambda collector function, which is triggered by an Amazon EventBridge Rule every night and collects the RSS Feed data and stores it in the S3 landing bucket
- AWS Glue table that pointed to the S3 landing bucket
- AWS Athena query engine that allows querying the data in S3
- AWS Glue Python shell job processes the data queries Athena and stores results in a tabular format in S3 for processed data.
You can find a detailed overview and deployment guide for this solution in my GitHub repository.
The data set is also publicly available and published here, so, you can explore it yourself.
Bird’s eye view
Over the last five months, Amazon Web Services made over 600 updates to 195 products, with 120 updates per month on average. Pretty impressive!
More than half of all updates were made to the services in six categories —
- artificial intelligence
- analytics
- databases
- management and governance
- compute
- security, identity, and compliance
I would expect to see the serverless and containers categories among those six. But it looks like this is just my personal feeling, and it has nothing to do with the actual status of things in Amazon.
Now, let us switch from the category view and see the products and services that received the most updates.
Amazon SageMaker, Amazon GovCloud US, Amazon Connect, and Amazon QuickSight are among the top five services and cloud regions that received the most changes! So, let us take a closer look at each service.
Amazon SageMaker — the most updated service
It is not a surprise to see Amazon SageMaker at the top of the list, as AWS continues investing heavily in extending machine learning capabilities, adding more features and algorithms, and simplifying model development and tuning.
Once promised to “put machine learning in the hands of every developer,” Amazon keeps its promise and continues evolving its AI ecosystem.
With 36 updates (45 if count updates to MLOps, Studio, Data Wrangler, and Canvas capabilities), Amazon SageMaker has become the most updated service in the Artificial Intelligence category and the entire AWS.
Amazon is trying to democratize and generalize ML and make it available for non-data scientists or ML professionals, by simplifying the machine learning process, automating model creation and evaluation, providing greater visibility into model predictions, and making the ML process faster, more accurate, and more robust.
Here are the most notable changes I would like to mention:
- Identify and limit bias in model predictions and training data with real-time explanations for ML predictions.
- Improved machine learning testing and deployment workflows by adding local testing to the SageMaker Pipelines (CI/CD for machine learning) and cross-account pipeline sharing.
- Improved accuracy and speed of ML predictions by introducing a new ‘Ensemble’ training mode for SageMaker Autopilot, reduced job start-up time by reusing training instances, and faster experiments using new hyperparameter initialization methods.
Amazon EC2
Amazon continues introducing new EC2 instance types, promoting the adoption of Graviton2-based instances, and making them available in different regions.
Also, EC2 is used as a foundation service for many managed services in AWS. So, it is expected to see more updates to one of the oldest services in AWS.
The most interesting changes are:
- Automated connection set-up between EC2 and RDS (Relational Database Service).
- 5 new AWS Controllers for Kubernetes (ACK) are generally available (with ACK you can manage AWS resources right from Kubernetes using kubectl).
AWS GovCloud (US)
AWS wants to secure itself as an official Cloud provider for the US government agencies, educational institutions, and other U.S. customers that run sensitive workloads in the cloud.
This region has received 31 updates in almost every product category, including Blockchain and Serverless. With most of the changes focused on Compute, Databases, and Security.
Amazon makes a steady effort to bring more services to the GovCloud US region, to make them, compliant and support stringent U.S. government security and compliance requirements.
Here are the most interesting changes that I would like to highlight:
- Amazon Lambda functions now support X-Ray tracing.
- Events from multiple regions can be combined into one central Region with Amazon EventBridge cross-Region routing.
- Added support for Amazon DynamoDB Standard Infrequent Access table class, which should lower storage costs for long-term storage of data that is infrequently accessed.
Amazon Connect
A cornerstone service for initiatives like Digital Front Door or Omnichannel Strategy, an Omnichannel Contact Center, Amazon Connect, had received 24 updates in the last five months.
It also became the most updated service in the Business Applications category.
Looks like Amazon sees many opportunities in the customer and patient experience segment. And wants to help customers streamline and customize the call center experience and provide easier integration with AWS.
The range of changes is wide. The most prominent of them are the following:
- Security improvements with voice spoofing detection during customer calls.
- The focus is on agent productivity (integration with Wisdom, schedule adherence). And visibility into the agent’s performance and bottlenecks (15 minutes scheduled reports, adherence, agent’s current activity).
- Personalization of the customer experience based on confidence scores and sentiment analysis.
- Automated merging of duplicated customer records based on confidence scores.
Amazon QuickSight
A data visualization service, Amazon QuickSight is at the top in the Analytics category.
This service has received 20 updates and closes the top five of the most updated service in AWS.
Multiple updates to Amazon QuickSight Q (a natural language query capability), region expansion, improvements to the UI and dashboard experience, and admin-level improvements to simplify and automate user and platform management.
Interesting new features:
- It is possible to identify questions that Q did not answer and enhance topic language coverage over time.
- You can now use Q Embedding SDK to ask questions programmatically.
- You can now embed QuickSight visualizations into your web site with fine-grained visual embedding.
Conclusion
It is interesting to see how AWS is changing and where they focus their attention. ML, Data, Compute, and Security are the main competencies that Amazon is developing right now.
Development in the US GovCloud region means more public sector companies will be using and utilizing cloud capabilities, providing better, secure services with a focus on customer experience.
Machine Learning will become more available for non-experts, so you only need to bring your data. The rest AWS will do for you.
Focus on staff performance and productivity managed by AI could become a thing soon, so we can expect more (AI managers or assistants) to appear in the near future.