Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
-
Curious as to why performance is your main concern? Serverless will always be slower than a server based application (unless you go for fargate). However that is not its main objective. Serverless is great for not caring about underlying stuff (scaling, patching, etc.) and just let you get on with writing business logic and composing applications from available services.
-
@dan-pud That's not true - it depends on the stack and the resources applied to your lambda function. A lambda function with a generous memory allowance (which also implies more CPU) is often way faster than a micro instance. We have benchmarks proving that's the case for part of our stack, I just haven't got it for node.js+dynamo specifically.
-
@AlmondSauce well it's for a small project ~6 months, once we have the PoC, we'd like to scale further whilst keeping the costs low. While lambda+dynamo is technically free i've noticed their post limits service is quite heavy with the costs. t1.micro is about 5$/mo give or take after the first year. Anyways, with the free tier in mind i just wanted to be aware of what i'm getting and how they compare with sheer perf and no other variables in mind.
-
@yellow-dog imma take a look at that, haven't considered going the sql route and i always thought PaaS was costlier than IaaS.
-
@dan-pud Well i have no real time experience with scaling, so having it handled automatically would be great but i also want to make the wiser choice with costs in mind, spending more hours on either isn't a bother for me since i'm a learner. So my decision lies where the money is easier, long-term setting in mind. If a simple t1.micro can handle more concurrent requests than the free tier lambda then the t1.micro would be my goto option, the bottleneck appears to be with the database though which is why i wanted to know mongo vs. dynamoDb(free tier) concurrent read/writes per second. I've seen people saying mongo supports 2500 ops/s (not sure how this plays out with t1.micro) which makes dynamodb's ~100 ops/s quite meager.
-
@AlmondSauce and what about the database, does dynamo limit this in any way (the ~100 free ops/second looks alright but lesser than the ~2500 ops/second a mongo database could offer)
maybe i could just use lambda + mongo on ec2. Then again if ec2 can't keep up with the requests it's the same problem all over again.
And the most useless function of the year award goes to...
Job interview
Ok so, i have no idea where i can ask this kinda thing so i'm asking it here (i know i could do like stackexchange or dead aws discord servers, ... nvm you know why i'm not going that route).
Anyways,
I'm looking for a comparison between a mongo+node setup on a basic t1.micro instance and a lambda+dynamodb setup.
Each one has it's perks obviously but i guess i sorta prefer whichever one gives best performance on the free tier.
I do know dynamo has 25 reads and 25 writes a second on the free tier, which might be a little less ? I really have no clue.
But how many writes/reads would a basic mongo setup be able to achieve on the t1.micro instance ? Any idea? Do share your experiences with these architectures as well. I'm sort of a newb with serverless, the downsides aren't worth it for me but I'm learning it nevertheless. It sorta tickles some sort of self-torture curiosity fetish (need more self-research to back that).
rant
grammar
db
aws
mongo
horrible
serverless