python - Kinesis-contributed latency using Lambda -


we're building (mostly) server less architecture centered around kinesis stream being consumed lambda, , running latency behavior don't understand , might block us.

we experiencing high latency @ extremely low volumes, set test case measure it: our lambda few things, logs, , posts message same kinesis stream trigger next invocation. test logs 100-200ms of billed time, invocations come 2-4s apart according cloud watch.

my reading suggests lambda should polling 200ms after function completeness, seems slow. found old posts eoy 2014 discussed similar behavior, appeared fixed in march 2015. expected? if not, there gotchas can result in slow invocation despite fast run times?

some notes: seeking <1s overall latency, optimally < 500ms contributed kinesis. using python library our lambda, did not have context.succeed() or context.done() methods, if might relevant. have experienced configuring greater memory, lower timeout, , different batch sizes no effect.


Comments

Popular posts from this blog

javascript - How to get current YouTube IDs via iMacros? -

c# - Maintaining a program folder in program files out of date? -

emulation - Android map show my location didn't work -