oeAware

oeAware is a framework for implementing low-load collection, sensing, and tuning on openEuler. It aims to intelligently enable tuning features after dynamically detecting system behaviors during system and service running. Traditional tuning features run independently and are statically enabled or disabled. oeAware divides the tuning process into three layers: collection, sensing, and tuning. Each layer is associated through subscription and is developed as plugins. oeAware can run in standalone or cluster mode. The tuning efficiency and scenario adaptation capability are improved using rules and heuristic methods, and the whole tuning process is transparent to service scenarios. Before and after tuning, performance is measured to check whether negative optimization exists. After a problem occurs, it is reported in a timely manner to suggest a rollback.

oeAware offers the following benefits:

oeAware enables tuning upon scenario awareness, and supports tuning recommendation, one-click enabling, and dynamic rollback. oeAware is a unified entry for tuning on openEuler. It unifies tuning consistency and change management.

• Improved management plane usability: Over time, static tuning can disperse tuning features of various scenarios across the system, making the tuning process difficult to manage and audit. oeAware resolves this challenge.
• Improved data plane flexibility: A single static tuning configuration can hardly adapt to all scenarios. Certain static tuning parameters may not achieve the desired outcome in complex environments. Additionally, hybrid scenarios can cause the tuning policy to fail, leading to higher manual analysis costs. oeAware also addresses this issue.