Optimize Memory Usage for Policies

[Pyatakov]

Problem Description

Currently, each policy consumes approximately 160 MB of memory. The policy service, which executes each policy in a separate thread, is the most memory-intensive service in our system. This excessive memory consumption limits scalability and increases infrastructure costs. Optimizing memory usage is critical to improving system efficiency and reducing operational expenses.

Requirements

• Analyze the memory usage of individual policies to identify optimization opportunities.
• Reduce the per-policy memory footprint while maintaining functionality.
• Optimize the policy execution model to use resources more efficiently.
• Evaluate and, if needed, implement alternative execution strategies (e.g., thread pooling, lazy loading, or policy instance reuse).
• Ensure that optimizations do not degrade system performance or introduce concurrency issues.

Definition of Done

• [ ] The memory consumption of a single policy is significantly reduced compared to the current 160 MB baseline.
• [ ] The overall memory footprint of the policy service is reduced without negatively impacting throughput or execution speed.
• [ ] No regressions in existing policy execution behavior.

Acceptance Criteria

• A single policy instance consumes significantly less than 160 MB of memory.
• The policy service’s total memory footprint is noticeably reduced.
• Tests confirm that execution time and throughput are not negatively affected.