I am experiencing severe latency issues and resource loading failures when accessing my Vercel-hosted website. This issue occurs only on specific ISPs in South Korea and affects multiple users across different devices. The website takes an unusually long time to load, and JavaScript and CSS assets fail to load intermittently.
After running a traceroute, I noticed significant packet loss beyond a certain point, suggesting a possible network routing issue or ISP-Peering problem affecting access to Vercel’s Edge Network.
Current Behavior (Problematic)
Very slow initial page load (~10+ seconds delay)
Static assets (JS, CSS, fonts) take excessive time to load or fail completely
Affects multiple users on the same ISP (Korea) but works fine on other networks
traceroute shows packet loss beyond hop 9
Command Used:
traceroute ganow.co.kr
Results:
traceroute to ganow.co.kr (76.76.21.21), 64 hops max, 40 byte packets
1 192.168.0.1 (192.168.0.1) 5.883 ms 2.050 ms 1.492 ms
2 221.147.62.254 (221.147.62.254) 4.423 ms * *
3 112.188.56.1 (112.188.56.1) 9.407 ms 4.180 ms 4.042 ms
4 112.188.53.177 (112.188.53.177) 4.004 ms
112.188.53.169 (112.188.53.169) 4.456 ms
112.188.52.205 (112.188.52.205) 3.661 ms
5 * * *
6 * * *
7 112.191.117.15 (112.191.117.15) 5.952 ms
112.191.117.33 (112.191.117.33) 6.505 ms
112.191.117.21 (112.191.117.21) 4.689 ms
8 112.191.118.189 (112.191.118.189) 7.729 ms
112.191.118.171 (112.191.118.171) 4.442 ms
112.191.118.165 (112.191.118.165) 4.411 ms
9 211.47.31.82 (211.47.31.82) 6.554 ms
211.47.31.66 (211.47.31.66) 6.551 ms 14.075 ms
10 * * *
11 * * *
12 * * *
13 * * *
14 * * *
15 * * *
16 * * *
17 * * *
18 * * *
19 * * *
20 * * *
The domain troubleshooting guide can help with most custom domain configuration issues. You might be able to use that guide to solve it before a human is available to help you. Then you can come back here and share the answer for bonus points.
You can also use v0 to narrow down the possibilities.
I am currently experiencing an issue where my Vercel deployment is being served from kix1::icn1.
Since my users are primarily based in South Korea, I would like my deployment to prioritize icn1 for better performance. I have already set "regions": ["icn1"] in my vercel.json, but the deployment is still being routed through kix1 first.
Could you please check if there is a way to ensure that icn1 is prioritized over kix1 for my deployment?
I have applied the solution you shared and confirmed that it works without any issues.
However, I am concerned about whether this method can be used permanently.
For now, it seems that we have no choice but to proceed with this setup, and we will need to keep monitoring it.
We have noticed that our application hosted on Vercel (https://ganow.co.kr) is currently being served from the kix1 region when accessed from KT ISP users in Korea.
However, for SKT and LG ISP users, it is being served from the icn1 region.
This issue has been persisting for the past two days, and we have consistently observed this routing behavior:
SKT / LG ISP → icn1 (Seoul/Incheon) (Expected)
KT ISP → kix1 (Japan/Osaka) (Unexpected)
Since icn1 (Seoul/Incheon) is geographically closer and should provide better performance for all Korean users, we would like to request a routing change to ensure that KT users are also served from icn1.
Many users in Korea are experiencing slow load times due to this issue, significantly affecting user experience.
As this is impacting a large number of users, we would appreciate it if you could resolve this as soon as possible.
Could you please update our Edge routing for KT users or advise on any possible solutions to enforce icn1 for all Korean users?
Looking forward to your urgent assistance on this matter.
When testing with vercel.json set to icn1 (Seoul/Incheon), the serverless function is correctly routed to Incheon. However, the middleware is still being routed to kix1 (Japan/Osaka), causing latency issues. This issue persists and needs to be resolved urgently as we are approaching the service launch. I would appreciate your help in resolving this quickly.
We confirmed that Vercel’s default domain (*.vercel.app) is being served from the icn1 region.
To resolve the issue, we redirected our custom domain to *.vercel.app using AWS Route 53.
This is a temporary workaround, and we hope that KT or Vercel can address the root cause of the issue.
I’m working as an engineer at a startup in Korea, and our company is also using the KT network. We are experiencing the same issue, and it is causing problems in our service operations. Please help us resolve this issue
I’m writing to let you know that this has been fixed as of 15:01 UTC on 2/14/2025. We’re sorry for the trouble this has caused you.
We have many edge locations across the globe, and we have multiple in South Korea as well. Specifically, we use AWS Global Accelerator to operate our Anycast network (related blog post)
On 2/7/2025, the AWS Global Accelerator team made routing optimization changes for edge locations in South Korea. This caused a subset of traffic from South Korea to be incorrectly routed to our kix1 (Osaka, Japan) region
This impacted a subset of traffic for ASN 9318, 4766, 17858, 9644, 17853, 3786, 9316
The issue was mitigated as of 15:01 UTC on 2/14/2025, as we made further changes to correct routing in South Korea.
Issue Summary
Current Behavior (Problematic)
(Expected)
(Unexpected)