- 最新
- 最多得票
- 最多評論
If we were only looking at network latency, 20 milliseconds is somewhere between 2,000 and 6,000 kilometres - that's a long way. But there are many other causes for latency. When you do a GET request there are (probably) load balancers, database requests, WAF rules, remote host CPU load - all sorts of things that are happening and many of them are going to be completely out of your control.
If you're sure that the remote server is in the same region I'd recommend trying different availability zones - it may make a difference (it also may not).
As a motherhood statement: Building an application around the lowest possible latency that is available means that it will fail or behave in unexpected ways when (not if!) the latency increases in the future due to some issue that (as above) is out of your control. Best to plan for higher latency than you're seeing. I understand that there are plenty of applications that depend on very low latency so finding the right combination of things to make that happen is important but there are laws of physics that can't be beaten.
相關內容
- 已提問 6 個月前
- AWS 官方已更新 1 年前