This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2022-04-14
Channels
- # aleph (2)
- # announcements (11)
- # aws (4)
- # babashka (42)
- # babashka-sci-dev (81)
- # beginners (90)
- # biff (2)
- # calva (40)
- # cider (16)
- # clj-kondo (26)
- # clj-on-windows (1)
- # cljdoc (4)
- # cljfx (1)
- # cljsrn (2)
- # clojure (92)
- # clojure-austin (2)
- # clojure-europe (23)
- # clojure-nl (5)
- # clojure-uk (3)
- # clojured (3)
- # clojurescript (19)
- # community-development (3)
- # conjure (1)
- # cursive (4)
- # datalevin (3)
- # datomic (5)
- # emacs (13)
- # events (1)
- # fulcro (26)
- # graphql (1)
- # hugsql (15)
- # introduce-yourself (5)
- # leiningen (1)
- # lsp (29)
- # minecraft (19)
- # music (1)
- # off-topic (36)
- # pathom (12)
- # podcasts (2)
- # portal (8)
- # re-frame (12)
- # reagent (11)
- # rewrite-clj (4)
- # shadow-cljs (56)
- # spacemacs (2)
- # vim (12)
- # windows (3)
- # xtdb (43)
Hi! Hope that everybody feeling good. Is anyone has a suggestion wether to use amazonica or aws-api (cognitect)? - note that we have to use presigned-url functionality which looks like it is still missing from aws-api, but maybe there are more things to consider. our mainly uses are S3, SQS, and cognito services. happy to hear your thoughts. thanks! 🙂
If you need presigned urls then you need the java sdk (which is surprisingly ok to use from Clj). Our internal library that simplifies working with AWS combines both aws-api and Java SDK (latter part is for IO intensive code or pre-signing)
I'd like to experiment on dynamodb to better understand its scaling behavior under very high load (e.g., 40k req/s). I have code written to run an experiment for some duration and to fire ddb GetItem or PutItem requests off at 40k/sec, spread uniformly over a single second. However, when running this with Cognitect's aws-api at 3k GetItem req/s, I will receive thousands of anomalies that look like this:
#:cognitect.anomalies{:category :cognitect.anomalies/busy,
:message "Ops limit reached: 64"}
So I'm curious, is the ops limit something I can change? Should I use the Java SDK for this instead? Am I just going to run into 1 compute limitations anyway?