Here’s your weekly breakdown of all things AWS…
AWS Serverless Application Model Gets Additional Features
AWS Serverless Application Model (shortened to AWS SAM) is AWS’ official answer to “how do I manage my Lambda functions?” It attempts to both standardize how Serverless applications declare themselves and provides a framework for Release Management that rides on the back of CloudFormation. Last week, AWS announced more features for managing their API Gateway service with AWS SAM. With these kind of features and the power of Lambda and API Gateway, a developer can build full HTTP applications completely serverless.
With this announcement, Amazon continues to lead the pack with “Serverless” architectures and demonstrates that they stand committed to supporting Serverless for the foreseeable future.
Amazon Enhances Their DNS Service Discovery Services
Earlier this year, AWS launched “Auto Naming” for their long serving DNS service, Route 53. Automatic Service Discovery for applications became a popular topic in the wake of Microservice architectures and Continuous Delivery. As more engineering teams migrate to CD or Microservices, DNS continues to serve as the most commonly used form of Service Discovery.
In their first offering, Route 53 Auto Naming did not support manually flagging a particular application instance as unhealthy. Earlier this week, that changed. Now, by using an API call or hooking into the AWS API with a third-party service, developers can manually choose to take certain instances in and out of rotation. This is handy for taking instances down for maintenance or for running more QA Automation and Automatic Tests on an instance before placing it in rotation.
Accessible Machine Learning
SageMaker is AWS’ managed Machine Learning service. With SageMaker, Data Scientists can quickly spin up an environment for training their models or even running their algorithms using fully integrated Jupyter notebook instances. There are even libraries available in the environment for doing things like data pre-processing using an Apache Spark cluster. Tensorflow and Apache MXNet are both available for training ML models.
Earlier this week, Amazon announced support for “lifecycle configuration” for notebook instances. This means that operators can run custom scripts on start-up, instantiation, or other notable events in start-up or tear-down phase of an instance. One of the most common use cases I see for this is to install and configure other dependencies like a Python Anaconda environment as soon as the notebook instance starts up. This will allow for easier customization for more flexible notebook environments.
CloudTrail + Athena Integration
I’m personally excited for this one. AWS just made querying over CloudTrail logs much easier. Using Athena, any CloudTrail formatted logs can now be quickly imported as Athena tables and ready for querying as if they were SQL tables.
Previously, this needed manual configuration but now it is as simple as pointing Athena at the S3 bucket where CloudTrail is storing the logs.
Teams can use this to quickly get useful access to their application and infrastructure events without importing them into a central logging system like Elasticsearch. I can imagine this will be useful for doing a fast, easy analysis of log data before importing that data into a more advanced event analysis tool like Splunk.
- Elasticsearch 6.2 is now available on Amazon Elasticsearch Service
- Amazon Aurora with PostgreSQL Compatibility is Available in the EU (London) Region
- CloudFront now Supports ECDSA Certificates for HTTPS Connections to Origins
- AWS CloudTrail Log Search Using Amazon Athena
- AWS AppSync now supports Android GraphQL subscriptions, releases sample photo application integrating AWS Amplify
- AWS Elemental MediaTailor Now Supports VPAID Standard and CloudWatch Metrics