How the cloud should work

There are lots of clouds. The one I'm thinking of is the one that Amazon runs, and that so many are trying to catch up to. Here's a list of ideas the next-layer-up will have.

  1. The basic unit shouldn't be a CPU, it should be an app. They are like the apps that run on your iPhone, but they run on a server, not on a hand-held device.

  2. I sign on to my account and see a list of my apps.

  3. To create a new app, click a button called New App. A dialog appears, asking what template I want to use (by template I mean GitHub repository, but that's something designers/programmers worry about). A configuration dialog appears, one created by the template designer. Checkboxes, text areas etc. These set the environment variables that configure the app.

  4. Storage is handled in a simplified S3-like store, without the quirks of S3. Static, web-accessible storage. Nicely configurable with user dialogs. Works like a file system and a web server. And it has an interactive mode that works like a file system (after all these years S3 still doesn't have one, because it's basically not possible).

  5. Double-click on an app to get a readout of what it's doing. Something like the Google Analytics dashboard pops up. I can see what kind of traffic it's getting. Look at how it's doing with its resources. Is a database filling up? Is response time okay? Has it been up continuously. This is what people who run server apps want to be able to see at a glance.

  6. If I want to add more resources, steal the slider from Heroku (see the video demo). The more resources the app uses, the more you pay. In times of peak load scale it up. When things quiet down, you can slide it back down.

This is what the simplified layer will look like. We can build so much more complex stuff when the basics that bog down deploying and maintaining servers gets simplified and commoditized.


Last built: Wed, Jul 9, 2014 at 11:23 AM

By Dave Winer, Thursday, April 17, 2014 at 9:19 AM. Good for the environment.