[Pycon] [new paper] "Marco Pavanelli" - Serverless or Containers?: two options to run python code in the cloud.

[info a pycon.it]

Title: Serverless or Containers?: two options to run python code in the cloud.
Duration: 45 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: Summary: we call them generically “containers” or Dockers and they are a very handy and popular way to put your software in the cloud while when we say “serverless” we mean an entire class of cloud products and services, in this talk we will see some examples and we will try to figure out what are main differences between them.   

First we will see some python code deployed in different popular serverless products, then some geeky things you can do with python using dockers and we will try to understand the strength and the limits these two approaches and finally we will see some cases where one one option is better than the other and why. 

Some very popular cloud services that are considered “serverless” are: Google App Engine (Standard) and Cloud Functions from big G, then we have Aws Lambda from Amazon in all its different shades and flavours and finally Azure functions from Microsoft. 
They all support reasonably well python, for some of them python language is the suggested choice, there are of course limits but when you are to decide where you will run your next cloud projects (I mean those you have not written yet) these are a good options. 

We will some examples of code running on these services and for each of them we will try to understand pros and cons. 

Dockers are popular for many good reasons: if you have an existing application running django 1.8 on python 2.7 and you want to replicate this running environment in cloud  safely and fast, dockers are certainly a very good option.
Dockers are very well supported on Amazon with different services like ECS and ECR, Google offers GKE with Istio and Microsoft offers AKS to help you manage your dockers in the cloud.  

With some examples we will see how simple it can be creating a cluster of dockers and how you can replicate easily the same cluster safely in the cloud. 


Tags: [u'#devops', u'#containers', u'#serverless']