Swami Sivasubramanian (AWS)
Amazon Web Services (AWS) has launched its machine learning (ML) service Amazon DevOps Guru, which can detect operational issues and recommend actions for remediation, into general availability
Utilising over 20 years of Amazon.com operational experience in building, scaling, and maintaining applications, the service uses ML to collect and analyse data, such as application metrics, logs, events, and traces for identifying behaviours that differ from “normal” patterns.
If something out of the ordinary is spotted that could point to outages or service disruptions, like increased latency, error rates and resource constraints, these issues are flagged through the Amazon Simple Notification Service and partner integrations, with AWS identifying Atlassian Opsgenie and PagerDuty as two examples.
Next, DevOps Guru estimates the potential impacts and causes of the issue, as well as “specific” recommendations for a fix.
“Customers continue to ask AWS for more services that enable them to take advantage of our decades of operational excellence in improving application availability running Amazon.com,” said Swami Sivasubramanian, vice president of Amazon ML at AWS.
“With Amazon DevOps Guru, we have taken that expertise and built specialised machine learning models to detect, troubleshoot, and prevent operational issues long before they impact customers and without dealing with cold starts each time an issue arises.”
In addition, the general availability also includes support for CloudWatch Agent Container Insights with coverage for Amazon Elastic Compute Cloud (EC2), Amazon Elastic Kubernetes Service (EKS) and Amazon Elastic Container Service (ECS), as well as an upgrade to its dashboard.
Amazon DevOps Guru’s general availability launch covers AWS’ Sydney and Singapore regions in Asia Pacific, as well as Tokyo, along with North Virginia and Ohio in its US East region, Oregon in its US West region, and Ireland, Frankfurt and Stockholm in its Europe region.
General availability is expected to expand to other regions soon, with AWS saying this is expected “in the coming months”.
The pricing for the service is currently the same across all regions. For the hourly cost of AWS resource analysis, the resource types are split into two pricing groups — AWS Lambda functions and Amazon S3 buckets are charged at US$0.0028 per resource hour, while other resource types are charged at US$0.0042 per hour.
In addition, DevOps Guru application programming interface (API) calls are charged at US$0.000040 per API call, or US$0.40 per 10,000 API calls.
