The main thing I see is observability. Nicer to lo...
# helpt
thdxr
03/23/2021, 1:12 PMThe main thing I see is observability. Nicer to look at performance stats broken down by function
f
Frank
03/23/2021, 4:37 PM
Yeah, heavily depends on the toolings u use. If u use xray to trace requests, it’s easier to glance through the trace if u had 1 lambda per request
Frank
03/23/2021, 4:42 PM
I always try to do both when I can’t make a call. In this case I actually have some of my Lambdas handling a bunch of requests. And eventually the one that suits my setup more would surface when I actually get annoyed by cold starts or feel like wasting time parsing thru the traces.
d
Dennis Dang
03/23/2021, 5:32 PMAre you using cloudwatch @thdxr or piping them logs over to something like datadog?
We use datadog (though haven't gotten it working yet in SST, WIP).
t
thdxr
03/23/2021, 5:35 PMYeah I guess I can be progressive about it and extract stuff as it makes sense instead of defaulting everything to independent lambdas
thdxr
03/23/2021, 5:35 PM@Dennis Dang just using cloudwatch right now but figuring out my setup
thdxr
03/23/2021, 5:35 PMmight try going through datadog since I've done that in the past but seeing if I can stay within AWS