- 最新
- 最多得票
- 最多評論
Is there a problem with the network environment from which you are accessing the site?
Not sure what kind of application, but how about checking the response time from access logs, etc.?
If you can configure the web server access log to record response time, you can check it.
If these results were similar for WA and QLD, I thought it was safe to assume that there was a problem with the access source.
Also, depending on the application, it may be possible to have CloudFront cache the data to speed up the response time.
Here I'd normally say the answer is latency - I'm extremely surprised that you're seeing lower latency in WA than in QLD. WA should be around 45-50 ms; QLD significantly less. Every round trip packet or transaction (at an application level) is greatly affected by the latency.
My rule of thumb is that for every additional millisecond of latency it takes an additional second to transfer one megabyte of data. That sounds over the top and for optimised protocols (like the AWS SDK communicating with S3) it's excessive; but for "chattier" protocols it is a pretty good measure. HTTP is normally ok but it can be offset by many connections to download small components - it takes time to establish each session; then to negotiate TLS; then to download.
As Riku mentioned: If you can use CloudFront and the content can be cached then it will absolutely help the WA users (there is a CloudFront point of present in Perth).
Another alternative is to put some of the application components in the Perth Local Zone. I've had customers see very low latency when using it; but it entirely depends on application architecture and whether it is easy to split up the components and run them in both Sydney (for the QLD users) and Perth (for the WA users).
We have AWS teams in both locations - reach out to them and also to the support team to help troubleshoot.
相關內容
- AWS 官方已更新 1 年前