Congested CDN / Routing Issue? [MY/SG]

Hi,

I am reaching out for assistance from the network engineering team regarding an ongoing issue involving frequent packet loss and elevated latency affecting connectivity via Fastly CDN.

The issue consistently manifests as packet loss ranging from 60% to 100% when routed through the following Singapore POP IP addresses, beginning approximately at 6:00 PM GMT+8 (10:00 AM GMT), progressively worsening from 9:00 PM GMT+8 (1:00 PM GMT) until around 11:00 PM GMT+8 (3:00 PM GMT):

  • 146.75.44.159
  • 199.232.44.159

Subsequently, around 11:00 PM GMT+8 (3:00 PM GMT), routing shifts to alternative POP IPs, resulting in latency increasing from an average of 20ms to approximately 90ms consistently for an additional 1 to 2 hours:

  • 146.75.112.159
  • 199.232.148.159

This pattern occurs daily with minor variations in timing, coinciding with peak local internet usage hours in Malaysia. Notably, despite the availability of a dedicated Kuala Lumpur POP, our traffic is consistently routed to Singapore-based POPs, indicating a possible routing inefficiency or ISP-specific routing decision.

Given the observed symptoms align closely with congestion patterns and considering local ISP capacity appears adequate, my initial assessment suggests potential congestion issues at the Fastly CDN Singapore POP.

Before escalating this matter formally with our ISP or local regulatory bodies—which could lead to significant consequences for the ISP—I would appreciate any insights or clarifications from Fastly regarding these routing and performance anomalies.

If you could send a request to an affected service (during the time when the problem is occuring) and include the Fastly-Debug header, that would be helpful. The details included in the response to that request can be sent to support@fastly.com and they’ll work with our network operations team to determine the cause.