Datomic client in AWS Lambda


Is it a bad idea to run Datomic client in AWS Lambda? I understand this is a problem for Datomic peer because of the large memory footprint.

I would appreciate if someone would hare some experience on working with Datomic client API within AWS Lambda.


Hi, have you tried using the Datomic Client in AWS Lambda yet?

I have been experimenting with it for a while now and the start up time is proving to be a significant hurdle. I’m using Clojure which will also add to the start up time, but from a completely cold start you are looking at around 30 seconds to get a response.

I’m looking at ways to mitigate this, by starting the service with a nice loader and then pinging it at regular intervals to keep it warm. However, AWS will from time to time start a fresh instance, even if there is a warm one, quite often when there is a lot of traffic, this new instance will then take around 30 seconds to return its response, leaving whatever part of the page requested the response hanging.

I really want to make AWS Lambda and Datomic Client work, but unless I can resolve these start up times, or at least ensure the user only has to deal with it at the most once per session, I really cannot see it becoming a viable solution.

I would also be very keen to hear from anyone who has got these two working together satisfactorily.

Thanks, Paul


Thanks for sharing your experience. I suspected start up time would be a problem but have not get around to try it yet. On top of my head, I can’t think of a solution either.