Technologies
AWS Bedrock
Amazon RDS for PostgreSQL
Amazon SageMaker Notebook Instances
AWS Lambda (SAM)
AWS Secret Manager
Amazon S3
You are opening our English language website. You can keep reading or switch to other languages.
Speeding Up DevOps & SRE Incidents Resolution with AWS Bedrock
Client and Challenge
Fiberplane sought to enrich their platform by integrating an engine capable of generating queries to Prometheus (a monitoring and metrics collecting tool). This feature aims to assist DevOps and SREs in gaining a clearer understanding of the most suitable labels to utilize.
Solution
To support this system, AWS Bedrock was utilized to enhance the natural language processing capabilities, while Amazon SageMaker Notebook Instances facilitated the development and training of machine learning models. The application relies on Amazon RDS for PostgreSQL to securely store and manage query data, with AWS Secret Manager ensuring the protection of sensitive information. AWS Lambda (SAM) was used to handle serverless compute tasks, and Amazon S3 provided scalable storage for datasets and model artifacts.
The main difficulty was thorough data preparation for the queries validation process in the complex multicomponent system. In order to overcome this, sources of possible supported requests were limited to four applications. Multistep filtering and subsequent improvement of the text descriptions for the queries were also applied in order to enhance the results and achieve higher accuracy of the final suggestions.
Outcomes
- Achieved 81% accuracy rate of the generated suggestions
- Natural language requests are converted into PromQL queries in order to retrieve environmental details from specialized applications
- Enhanced code completion accuracy
- Improved operational efficiency and faster code development
Contact UsPlease provide your contact details, and we will get back to you promptly.
