Amazon Web Services (AWS) has unveiled Amazon DevOps Guru for Serverless, a service that uses machine learning to improve the operational availability and performance of AWS Lambda serverless applications.
Introduced April 21, the AWS Lambda support is a new feature of the Amazon DevOps Guru service for monitoring application behaviours. Amazon DevOps Guru is also available for all Amazon Relational Database Services.
Amazon DevOps Guru uses machine learning models informed by years of AWS and Amazon.com operations to help developers improve application performance.
Developers using AWS Lambda can use the service to automatically detect anomalous behaviour at the function level and use ML-powered recommendations to remediate any found issues. Problems can be detected such as underutilisation of memory or low-provisioned concurrency.
When an issue is detected, Amazon DevOps Guru for Severless displays findings in the Devops Guru console and sends notifications via Amazon EventBridge or Amazon Simple Notification Service (SNS). To get started, developers can navigate the DevOps Guru console to enable the service for Lambda-based applications, other supported resources, or an entire account.
Specific operational issues and proactive insights available from Amazon DevOps Guru include AWS Lambda concurrent executions reaching account limit, triggered when concurrent executions reach an account limit for a continuous period.
This is in addition to AWS Lambda provisioned concurrency function limit being breached, set off when the reserve amount of provisioned concurrency is insufficient over a period. Also, AWS Lambda timeout is high compared to Simple Queue Service's visibility timeout, triggered when the duration of the Lambda function exceeds the visibility timeout for the event source Amazon Simple Queue Service (Amazon SQS).
Delving deeper, account read/write capacity for Amazon DynamoDB consumption is reaching the account limit while AWS Lambda provisioned concurrency usage is lower than expected and Amazon DynamoDB table consumed capacity is reaching the AutoScaling Max parameter limit.