Service Resource Optimization Strategy Guide
Background
As user numbers grow and feature demands increase, we have observed that certain high-resource-consumption configuration options can lead to service instability. To ensure service quality, we conducted an in-depth analysis and formulated a corresponding optimization plan.
Resource Optimization Strategy
1. Filter Update Mechanism Optimization
Current Situation
- Some users have configured hourly filter updates
- Each update requires a full download-parse-deduplication cycle
- International bandwidth limitations prolong update times
- Server resources remain under continuous high load
Optimization Plan
We have adjusted the minimum update interval to 72 hours for the following reasons:
- Most filter lists update on a 24–72 hour cycle
- Reduces ineffective resource consumption
- Ensures service stability
- Improves bandwidth utilization efficiency
Impact Assessment
- Positive impacts
- More stable service response
- More reasonable resource usage
- Reduced system load
- Minimal impact
- Rule updates still occur within a reasonable cycle
- No effect on protection efficacy
2. Parallel Request Strategy
Current Situation
Most users currently have parallel requests enabled, but under the existing architecture the benefits are limited:
- Latency differences among Alibaba Cloud upstream services are typically within 5 ms
- May trigger Alibaba Cloud public service request-rate limits
- Adds unnecessary system overhead
Usage Recommendations
- We recommend using load-balancing mode
- Parallel requests are suitable for:
- Scenarios with significant upstream latency differences (>200 ms)
- Cases of unstable service quality
- Cross-border access scenarios
Note: We have not yet observed any throttling issues caused by parallel requests, so this feature remains available for now.
3. Third-Party List Management
Security Considerations
To ensure system stability, we have temporarily disabled support for certain third-party lists:
- External list sizes are unpredictable
- May lead to resource overruns
- Service stability cannot be guaranteed
Future Plans
We are researching a safer third-party list management solution so that this feature can be re-enabled in the future.