0

I need to change configuration values without requiring deployment of code changes. I had the following doubts related to using AWS AppConfig to confirm it is applicable for my use case.

  1. What actually happens in an AWS AppConfig deployment? Will the AWS AppConfig deployment send the newly updated configuration values to each host of the onboarding service and now the service hosts can just pick up the values without depending on any remote hosts? Or is it like the AWS AppConfig deployment sends the newly updated configuration values to remote hosts of AWS AppConfig and for each service request, the service host needs to call the remote AWS AppConfig remote host?
  2. Is caching available out of the box? By caching, is it like the service host calls the AWS AppConfig hosts only if its cache doesn't have requested value already or something else like the service host always calls AWS AppConfig hosts but AWS AppConfig hosts only call underlying database after checking their cache first?
  3. What is the maximum traffic per second that AWS AppConfig can support?
  4. What is the maximum latency to get the configuration from AWS AppConfig?

Thanks

Gopikrishna K S
  • 129
  • 2
  • 7

1 Answers1

1

Good questions. Here are my responses. (Full disclosure - I work on AWS AppConfig.)

  1. What happens with an AppConfig deployment? Answer: AppConfig will deploy the new configuration to your targets according to the Deployment Strategy you specify. Your hosts will poll AppConfig for new data on the intervals that you have set up. If you have a gradual deployment, each host will make independent calls to AppConfig, and will pick up the new configuration data in the time and percentage that is specified in that Deployment Strategy (minutes, hours).
  2. Caching client? The only current out-of-the-box caching client is for Lambda users, using the AppConfig Lambda Extension. Otherwise you will need to handle your own caching logic.
  3. TPS Limits? The current limit is 1000TPS.
  4. What is the Max Latency? Latency depends on a lot of factors, including where your hosts are and where you are calling. There are not published max latency numbers, so you will have to try in your set up. But of course we try to reduce latency to very low. Caching config helps with this too.

I hope this helps.

Steve Rice
  • 36
  • 1
  • To 2-4, AppConfig is designed for you to poll every X seconds and then cache on your end. Each call tells you how long to wait until the next call. Asking about TPS/Latency makes me believe that you're (OP) planning on calling this during a request, which is not the intended use model and a really quick way to run up a gigantic bill. – Tyler Carter Sep 14 '22 at 15:17