Performance degradation on Lightsail Tokyo instance during peak hours (ap-northeast-1)

The performance degradation you're experiencing on your Lightsail instance in the Tokyo region during peak hours is likely related to the burstable performance model that Lightsail uses.

Lightsail instances operate with a baseline CPU performance and can burst above that baseline for limited periods. Your instance has both a sustainable zone and a burstable zone for CPU utilization. When your instance operates in the burstable zone (consuming more CPU cycles than the baseline allows), it depletes its burst capacity. Once these burst credits are exhausted, performance drops to the baseline level, which can cause significant slowdowns.

The timing pattern you've observed (poor performance during peak hours and recovery late at night) suggests that:

1. During peak hours (20:00-23:00 JST), your instance may be consuming more resources due to higher traffic or workload, depleting its burst capacity.

2. During off-peak hours (after 01:00-02:00 JST), the lower demand allows your instance to recover its burst capacity, resulting in better performance.

The dramatic difference in download speeds (1.65 Mbps vs. 241 Mbps) is consistent with what happens when an instance exhausts its burst capacity and falls back to baseline performance.

Regarding your specific questions:

1. While there isn't specific information about known peak-hour network congestion in ap-northeast-1, the pattern you're seeing is more consistent with burst capacity depletion than regional network issues, especially since your upload speeds remain relatively high.

2. Yes, burstable instance types can experience significant performance degradation during periods of high usage when they exhaust their burst credits.

To address this issue, you could:

• Upgrade to a larger Lightsail plan with higher baseline performance and more burst capacity
• Optimize your application to reduce resource consumption
• Monitor your CPU utilization and burst capacity using Lightsail's monitoring tools
• Consider scheduling resource-intensive tasks during off-peak hours
• If consistent high performance is required, you might want to consider migrating to EC2 which offers more predictable performance options

If the issue persists after trying these solutions, it would be worth contacting AWS Support for more detailed investigation.
Sources
Identify when your Lightsail instance bursts - Amazon Lightsail
My website hosted on Lightsail goes offline at different times every day. | AWS re:Post
My LightSail Bitnami LAMP is too slow and at times unresponsive | AWS re:Post