Hey i have question here, Things that we can do inside a container ( let's say hosting a website) we can't do this in aws lambda right And lots of other stuff too ?
Hey Digvijay, the architecture changes a bit. The static website will be hosted with s3 and CloudFront and dynamic API calls will be done using API Gateway Lambda, and you gain high availability and scalability out of the box. Check out my Netflix/prime system design video for this architecture, this part is near the end(timestamp included in that video).
Thanks for the video. But I just wanna point out that the distinction between serverless and container is not applicable to GCP. In the GCP world, you can have a serverless container via Cloud Run.
My Kubernetes on EKS Course is out now in Udemy!! If interested, please check out discounted ($9.99) link - www.udemy.com/course/rocking-kubernetes-with-amazon-eks-fargate-and-devops/?couponCode=JUN21BP2
It was enlightening! You can differentiate two things only when you know about those two in depth. This proves your concepts are rock solid! Now the Concept is clear for us too. Many thanks Rajdeep!
Ty Marc for the kind words! I am so happy that you got something out of my videos. If you like Serverless, feel free to check out the other serverless videos in my channel. Thanks!
Depends Garrett. One of the reason companies adopt serverless is pay per use. With serverless running on top of container (for example using Knative), you still pay for the underlying worker nodes even if they are underutilized. There are other factors such as rate of scaling, service integration etc. But yes, this is definitely an interesting concept!
So to scale containers you need to scale VM? I thought that with containers you can just scale and run more containers from the same image without creating more VM. Or am I misunderstanding something?
The pods wrapped around container(s) runs inside a node (EC2). When the traffic goes up, HPA (Horizontal Pod Autoscaling) will kick in and create more pods in same node(EC2). However when demand increases and current node(EC2) has no capacity left, it will engage cluster autoscaler and another node (VM/EC2) will be provisioned and new pod will go there. Hope this helps. Thanks for the question.
What is serverless computing? Like PaaS, serverless computing enables developers to build applications faster by eliminating the need for them to manage infrastructure. With serverless applications, the cloud service provider automatically provisions, scales, and manages the infrastructure required to run the code. Serverless architectures are highly scalable and event-driven, only using resources when a specific function or trigger occurs. It's important to note that servers are still running the code. The “serverless” name comes from the fact that the tasks associated with infrastructure provisioning and management are invisible to the developer. This approach enables developers to increase their focus on the business logic, and deliver more value to the core of the business. Serverless computing helps teams increase their productivity and bring products to market faster, and it allows organizations to better optimize resources and stay focused on innovation.
in the 90Million compariosn , the different is ( 206 - 173 ) = 33$ is just infra structure cost but not the total Operational cost , we need Manpower for support and Admininstration of cluster which is not in case of Lambda . where this additional cost is justified is when you said the brownfield apps that can not run on lambda Containers are the best Option . what are your thoughts on this ?
Lambda examples are in python. However most lectures are about concepts and not specific to any programming language. Note that Serverless is NOT just lambda! Course covers AWS Serverless Architecture, Lambda, API Gateway, DevOps, CI/CD, DynamoDB, SAM, Cloud9, Serverless Frameworks etc.
13:00 - in your model must to take in consideration the complexit of the system. If you use containers must to pay for extra HR to manage servers. The model is nor 100% accurate but it is a good start point. Thanks
At 13:20 when you are showcasing lower cost for Container approach, we do need to take the cost of resources for managing the Kubernetes clusters/nodes. rit ?
0:22 i can't explain how much i appreciate simple note taking like this. some pros keep showing their damn face all the time and it really sucks for the learner. thanks for your effort . it is perfect
You are absolutely right, it is NOT a TCO comparison. I kept it out coz the cost of TCO varies based on many factors. I just wanted to deal with the definite cost for the youtube video. When I do real life cost comparison for enterprise, we do take into account datacenter power input, maintenance cycle, number of ppl maintaining project, doing management task etc. Good point peng, thanks!