I'm managing a gaming app backend on AWS in the eu-west-2 region, and the API latency is frustrating for users located far away. For instance, players in London are experiencing latency of about 100ms, while those in Malta are seeing about 200ms. However, users in New Zealand are facing much higher latency, ranging from 700ms to 1000ms. I've tried various solutions, but nothing seems to work:
1. CloudFront caused issues with our authentication when it modified requests.
2. The Global Accelerator didn't support SSL termination.
3. Using Cloudflare with Argo still resulted in high latency of over 700ms.
4. Combining Cloudflare with Global Accelerator and ALB didn't yield any improvements.
I can't go for a multi-region setup due to compliance and data regulations. Is this high latency just a result of the physical distance from New Zealand to London, or is there something I'm overlooking? What do other gaming platforms do to address this issue?
2 Answers
Many have tried to solve the issue of latency from long distances. The simplest methods are:
- Physically relocating your servers closer to users when possible.
- Utilizing UDP for latency-sensitive components.
- Splitting your backend logic to focus on real-time needs while skipping heavier services like Cloudflare for those critical paths.
Unfortunately, you can't escape the speed of light! The best approach to improve user experience is to have your service run in multiple regions. This way, players can connect to the closest server, significantly reducing latency.
I haven't looked into GameLift yet—does it really just act as a wrapper for EC2? I'll check it out!