As part of its last-year-old customer event re: invent amazon had already introduced the service codeguru. Now developers are the tool for quality control and code optimization released in a version (ga – general availability) available for productive use. Codeguru uses the analysis of machine-learning algorithms to create proposals for more efficient codes design and contribute to a higher quality.
The tool adapted to use with a variety of aws services combines two components, a reviewer and a profiler. Codeguru reviewer can integrate developers like other prousing tools in their development process when they import new java code into their repository. Codeguru not only support aws codecommit, but also github, github enterprise or bitbucket cloud. Codeguru reviewer then open a pull request and starts checking automatically. The integrated ml models have been amazon with code review data of more than 10.000 open source projects trained on github.
Check and examine
The second component, the codeguru profiler, should help developers to find their "most expensive lines code", as it is called in the amazon blog post to envision of the tool. For this purpose, the profiler developers gives insights into the runtime behavior of their software to provide any performance bottlenecks and to initiate measures to resolve as well as the general optimization of the code. The aim is to improve the java code (or more jvm languages) so far that the compute costs for use of the software on cloud services such as aws lambda, amazon ec2, amazon ecs, aws fargate or aws elastic beanstalk, but also on-premises, have to be kept low.
The paid ml service amazon codeguru is now available to all aws users. After a maximum of 90-day testing, costs of 0.75 us dollars per 100 code lines as well as 0.5 cents per sampling hour and application profile. Clear information about the tool can be found in the envision or on the amazon codeguru website.