Cloud Development Kit (CDK) by AWS will completely change software
Modern infrastructure has been done via configuration files (cloudformation, ansible, salt stack, chef, docker), allowing the automation of the creation and scaling of infrastructure. This was once a major innovation.
CDK by AWS ups the level of abstraction, allowing for one to create object oriented abstractions over their complete software infrastructure. This, in the next five years, will completely change software development. We have never seen this level of abstraction, with the whole power of the AWS stack at the finger tips of the developer. How can a team justify not being on AWS?
CDK is to cloudformation as c++ is to assembly.
Templates of complete software stacks, with all devops built in, will be created and will completely automate dev ops. Whole websites can be spun up in minutes, pieces of them mixed and match with open source infrastructure templates pulled in.
Imagine a spring package where your ORM already has an aws database baked in, with the automated switching between dev and prod, profiling, etc. Right out of the box. Just a random example, but the automation here is massive
Why am I not reading more about this?
29 comments
[ 2.6 ms ] story [ 68.2 ms ] threadMore and more people are dying from heartattacks caused by understanding the true costs of losing control and relying on cloud religion and throwing money into Bezos' fireplace.
You can do the exact same thing with Pulumi and it's cross platform.
Also, in my experience, CDK is an absolute nightmare if you like running things locally and testing on your own machine
To add to this, Pulumi has a few features not yet present in CDK like encrypted state, and dynamic resources. The TLS/Random provider also seems to be missing from CDK but present in Pulumi. I am using all of these features in some Pulumi projects and couldn't imagine using CDK, ever.
tl;dr Use your preferred tooling, but don't assume a team couldn't justify using their tools of preference.
Ciagenix is a big scam. So if you are going to put your money on this product; you are not going to get value for your money. There are so many things wrong about it and any sane person would easily tell this product is big scam. Take this for example; this formula has not been put through any clinical test. So we are not even sure; even if is safe or even effective.A lot of common health issues have been reported by some few people who have tried it before. The formula come associated with some serious side effects including; headaches, insomnia, racing heart and also feeling like you are going to pass out.
http://supplementstore4u.com/ciagenix-review/
https://sites.google.com/site/supplementstore4ucom/ciagenix-...
https://publons.com/researcher/4145675/ciagenix-1-sexual-pil...
google is better
especially docker and other infrastructure via bazel.
the interesting bits in-development these days are not the toolkits to assemble cloud. the interesting bits are the cloud-ification of the application. rather than configuration & state being managed by the app, externalizing more & more of the inner-pieces of the app into external services really reduces the app foot-print, & gives all apps in the environment a common basis for operation. at the moment, it's pretty much only kubernetes doing anything here, & frankly, it's under-hyped, radically under-hyped, that this is the true-er goal of kubernetes. it's not just another platform. desired state management is a whole philosophy for computing, about writing what you want to do in to a central archive, then allowing autonomic systems to break down & accomplish the state you have told the system to set out for.
this hinges upon simple steps like externalizing your apps state into central/cloud systems along-side other app's state. it's practiced by writing separate controllers, a kind of inten-based CQRS 2.0. writing code to build state, as cdk does, is convenient & ops friendly. but still very procedural/computing 1.0. i very much look forward to a far better, far more expansive future, one that is far less isolated "me and my app" & much more bigger picture & all encompassing. kubernetes is bringing that. nothing else is, that i have seen.
https://www.pulumi.com/docs/intro/concepts/programming-model...
https://news.ycombinator.com/item?id=25641205
regarding privacy in Google, to be downvoted almost immediately. There's people that truly don't care about privacy and control, and see comments like this as a nuisance, because they can't see how this massive amount of information could be used against them. WEF itself has been selling the you won't own a thing, and you'll be happy mantra for a while, and there's more and more people buying it.
It would be palatable if the ones promoting it didn't have so much privacy and control themselves.
The proposed model for the future is obviously abstraction with same patterns in hand. People will buy it to some extent I guess but the idea of ownership is hardwired to our psychology naturally, ownership is connected to privacy in direct way, I cannot see how Bezos, Gates, Musk will comply with this. What will they do? Will give away all the money they have and live free of ownership? GTFO.:)
I've recently started to merge IaC code with runtime code, so that Lambda functions and AWS CDK projects share the same dependencies and the same package.json. I think Pulumi does something similar, although I'm not familiar with it. It makes development one step simpler, because you only have a single set of dependencies to worry about in a project. Esbuild takes care of only deploying the code you actually need to the Lambda functions.
The next step I hope to see is TypeScript type support for AWS Step Functions. Currently there's no type checking for the input and output objects of Step Function States, or for the various selectors you can specify for them. With proper typings, AWS CDK could throw deploy-time errors when Step Functions are badly configured or don't match the input/output types of related Lambda functions. Actually, you might see the problems already while editing the project in Visual Studio Code.
The future direction here is that you can implement fully featured cloud applications, including runtime code, with very little boilerplate code, just by defining data structures, essential business logic and high-level cloud resources. AWS CDK will take care of all the details by default.
We already know of a few cases where a bug in code got a few thousands of dollars of damage.
Imagine that multiplied by the number of objects created on an array that generates clouds in real-time in a developer's code.
It's a pretty pricy bug.
I can see a future where only 1 cloud remains and all our tools and resources are on the Cloud Provider's Infrastructure and we cannot control or even self-host our own tools and data and everything we do and host is being monitored. If someone does not agree with what we store or host, it may be rejected and even banned/deleted WITH our consent (because we accepted the EULA) but without our authorization. We would not own anything we create or send to the cloud.
We would all become Cloud Developers, Cloud Web Developers, Cloud Engineer, Cloud SRE in our resumees.
This should be more open and not be driven by Vendor Lock-in.