AWS Lambda Provisioned Concurrency | Lambda Scaling and Concurrency 101 | Lambda Account Concurrency

Learn how AWS Lambda Scale and achieve concurrency. Learn Lambda Account Concurrency and Provisioned Concurrency. Also what are your holiday goal?
Provisioned Concurrency Blog: aws.amazon.com/blogs/aws/new-...
Connect with me in LinkedIN: / rajdeep-sa-at-aws
My Serverless Course on Udemy: www.udemy.com/course/aws-serv...
Definitely do NOT Click: / @cloudwithraj

[@pokmnhyu]

Great Video Raj !

[@shwetashaw9978]

Simple and crisp👍

[@zackriso3945]

Hi Raj, I read in multiple articles that Lambda runs on Firecracker, which is an LVM, and which runs on EC2 as a backbone, rather than containers (ex. ECS). I am referring to the underlying infrastructure behind the scenes, not the possibility to invoke lambda through an ECR-Container (instead of a Zip file).

[@lzoneonethreethreeseven8046]

are lambda functions always scaled automatically when many requests come in? have lambda functions an integrated load balancer or is it vital to provide one that targets it in every case?

[@csnayak]

awesome explanation

[@amitk5088]

superb sir my problem resolved I make it 200 now it is working

[@cloudwithraj]

Great 👍

[@neerajchouksey3761]

Thanks !!

[@karan_1993]

Hi Raj, your content is super awesome. Plz keep on making more such videos.

[@cloudwithraj]

Ty Karan for the kind words

[@namratacareer]

I am hitting an lambda API from react code repeatedly, so the problem is the sequence I am sending the request ,response are not coming sequentially, response of recent request comes faster than my previous request. So how to configure throttle to handle this scnerio

[@Cedielagekohaha]

If you want to handle the request sequentially then you need to use FIFO queues backed by Lambda. Regarding throttling, you can configure your queues to have a redrive policy (retry) whenever the request fails. If the maximum redrive policy is reached it will be rerouted to DLQ .

[@rasensio]

Great explanation :)

[@cloudwithraj]

Glad you liked it!

[@anhquocnguyen1578]

your explanation is very well. keep up your amazing work, pls. thank you

[@cloudwithraj]

Thanks a lot!

[@75neelshah]

that is good explanation what is difference between provisioned concurrency and snap start

[@amitk5088]

Thanks a lot sir

[@cloudwithraj]

Most welcome

[@as_wani]

By default the unreserved account concurrency is 1000 and if I raise a request to change it. say I want it to be increased to 2000, AWS is going to charge me for that or not? If I reserve concurrency for one lambda function to 200, does it mean that function is always going to use that limit and never go beyond that limit?

[@cloudwithraj]

You will NOT get charged if you just increase the limit. You only gets charged on what you use. If you allocate concurrency limit for a lambda to 200, it will never have more than 200 concurrent executions (Your assumption is right for this one). Thanks for watching!

[@cloudwithraj]

@Shanmukh Meesala You will get a throttling error (error code 429), and you can retry.

[@galeop]

"Reserved concurrency" therefore means "max concurrency". "provisioned concurrency" means "number of always running instances"

[@cloudwithraj]

Correct!

[@galeop]

thanks!

[@biswajeetsethi7689]

Hey Sir. I understood how Lambda gets the container that is behind the scene. But if we go to the system level. Who provides container service to Lambda ???

[@cloudwithraj]

Hey Biswajeet, AWS firecracker provides the underlying container. Now, you can also create your own container - kzbin.info/www/bejne/fn_QZ52LlKaImKc

[@amitk5088]

can you help me

[@takethree1]

I have 30k SQS msg/min for Lambda to process. Do I need provisioned concurrency? increased ?

[@kavyasuresh7423]

Did you solved this problem?

[@amitk5088]

I am getting this issue

[@ziadirida]

Isn't it true that lambda is great because you pay nothing with serverless when you are not using it. But with provisioning, this advantage goes away.

[@Conlexio]

there is also an advantage for simpler projects in that you don't have to manage and maintain servers too I guess

[@cloudwithraj]

Yes, you do pay extra for provisioned concurrency. Best practice is use provisioned concurrency based on schdule if you know traffic pattern, or you can make it scalable based on traffic.

[@Cedielagekohaha]

Everything depends on the application use-case. If you have strict response latency SLA then having a provisioned concurrency is great.

[@amitk5088]

Function is throttled Your function cannot be invoked while the reserved concurrency is zero.

[@cloudwithraj]

Your function has no concurrency left. Either all other lambdas are using the concurrency available in the region/account or the concurrency of the lambda in question is set to zero. Change it to some positive value and then retry

[@amitk5088]

@@cloudwithraj yes sir and I am going to buy today your Udemy course of lambda. I am from India sir thanks a lot sir.......

[@amitk5088]

@@cloudwithraj finally today I buy your course from Udemy, Now I am going to do practice....thanks a lot sir I got this [e-mail] my email id - [e-mail]

[@amitk5088]

Test Event Name delhicode Response { "statusCode": 200, "body": "\"Hello from Lambda!\"" }

[@ujjwalh8000]

Whats with the fake accent?

[@mrragamuffin]

New York