Deployment Models

Sahil Malik: Now let’s talk about deployment models for Azure Functions. Azure Functions supports continuous deployment. As the name suggests, if you check in code into some sort of source control, Azure Functions supports many of them. You can do continuous deployment directly from your source code onto an Azure Functions. I’ll show you how this works momentarily.

You also have the option of Zip deployment where you upload a zip file and the files get unzipped and they get placed in the wwroot folder, or you can run directly from a package. The main difference between running from a package and a zip folder is that various deployment methods, including the zip deployment, rely on copying files to the wwroot folder, and after the copy’s done, that runtime takes over. So it’s no longer relevant how the file scarred there, but as long as the files are there, when the run time takes over, things start working.

So the runtime then takes over and it runs a logic based on whatever is in your files. But with run from packaged, things are a little bit different. There is no longer a deployment step which copies individual files to the wwroot. Instead you pointed to a zip file and the zip file it gets mounted at wwroot as a read only file system. In other words, you have a lot of advantages here like fast deployment, atomicity, core start performance, versioning, et cetera, right? And this currently doesn’t work for on Linux but offer other options. A run from package is a very good choice. Then you have automated resource deployment where you can use an Azure resource manager template to deploy a function app in a script of old fashion. And finally, on premises functions currently under preview, allow you to run the Azure Functions around time on your own premises iOS servers.