CloudConf 2016

Posted on by Gianluca

Edit: despite the time passed since I wrote this article it seems that there is still someone that is finding it interesting and entertaining! I’m really happy about it and so, thanks a lot Eric! 😄

Regarding interesting posts let me grab the chance to suggest this one wrote by James Crace of Cloudwards 👌.
Go on and take a look at it, I’ll wait you here don’t worry 😉

…oh there you are, welcome back! Ejoy my post! 😁

Last 10th March I had the pleasure to participate to the fouth CloudConf. This year the conference was hosted in Turin, in the beautiful “Centro Congressi Lingotto”.

From its beginning in 2013 the conference has gained a lot of traction thanks to the level of the talks and its great organization.

This year in particular the event took the form of a one-day conference between a two-days workshop session. Unfortunately I had the possibility to attend only the conference but the workshops seemed particularly intersting as well.
The topics addressed were specifically:

  • the use of the PHP language in the context of the Cloud,
  • Amazon Web Services,
  • Docker and the serverless principles and architectures.

These topics represent also some of the themes covered during the conference. But beware! Don’t get fooled by its name!

The topics covered were indeed mainly pertaining the “Cloud” and the DevOps culture but they also encompassed themes like IoT and both software and data modeling.

In general a lot of focus was placed on development automation, serverless infrastructure, cloud services, testing infrastructure and IoT architectures. To have a complete overview of the topics treated here there is the schedule of the day.

Personally I followed all the talks that were held in the big hall, i.e. “Sala 500”:

During the morning the “Sala 500” hosted five keynotes where we were presented various experiences and services by important representatitves of companies like Altassian, Amazon (Web Services), Docker, Google and Elastic.

In “Rise of the machines: Automate your development” Sven Peters talked about all the automated solutions used in Atlassian to simplify their daily work.

Some examples are the automation of stand ups, the control of issue tracking through GitHub commits and user actions, the autosuggestion of most relevant reviewers and much more. All this is accomplished thanks to Atlassian self developed bots like Agent Charlie, Dr Code, Halleluya, Flaky, and the most known Hercules and many more.

Next was Danilo Poccia from AWS with “Building event-drive serverless applications”. In its presentation he detailed many features of the AWS Lambda service and its potential concerning the development of serveless architectures. To conclude he also delivered a live demo regarding many of the concepts he exposed during his talk.

Right after a little break it was the turn of Aanand Prasad, main contributor of Docker Compose previously known as Fig. With his “The Internet is a computer” he beautifully portraied many basic but powerfull concepts regarding the need to abstract away complexity and simplify the life of other developers. This was (and still is) one of the main Docker Compose objectives.

Just like Danilo, he also delivered a demo in which he showed the potential of another technology he’s working on, i.e. Docker Swarm. He showed how it can ease the handling of clusters of container by considering them like single entities and abstract away their inner complexity.

Following the same line of thought, Alexis Moussine-Pouchkine from Google, presented with his “Google Cloud Vision API: what a great time to be a developer!”, some of the key priciples followed by the “big G”, like “there is no “Big Data”, there is just “data”“, “containers over hosts”, “autoscalability everywhere” and “WYGIWYS”. Together with this he also reported how Google use Machine Learning (ML) to empowers many of its services. Google Search (also the API), Google Translator, Gmail spam filters, Google Photos, are just a few names. Others, maybe not well known, are the Tensor Flow, a library specifically designed for ML experts and the brand new (still in beta) Google Vision API.

The morning schedule ended with “Awesome loggin infrastructure using the Elastic stack” from Alexander Reelsen of Elastic. In his talk he portraied a general overview of virtually all the services provided by Elastic that can help one of the most important task in the context of software development and infrastructure management, i.e. logging.

As previously said, in the afternoon I followed four of the six sessions held in the “500” hall.

The first one was from Mauro Servienti of Particular Software. In his “The road to a service oriented architecture (SOA)” he focused on explaining the main concepts behind SOA and the main misconceptions regarding this architectural pattern. He focused in particular on the importance of messages and asynchronous message passing by presenting some practical use cases.

Next on stage was Oded Coster from Stack Overflow with his “Stack Overflow behind the scenes how it’s made”. After presenting a few astonishing numbers regarding the performance of Stack Oveflow he revelead the main buiding blocks and technologies constituting their infrastructure. And it is nothing like you may imagine. Bare metal with Windows stack: nine IIS Web Servers, four SQL Servers, one Redis Server, HAProxy, Elastic Search and Tag Engine.
How they achieved their results with these assets?
Simple: consider performance as a first class feature of the application and apply the scientfic method to improve it at every possible levels (even hardware!). The Cloud? Well it didn’t fit their needs!

From Stack Overflow bare metal to one of the most notorious technologies of the moment: Docker!
With “Docker Advanced features” of Antonio Murdaca] we all had the possibility to discover some new hidden features introduced with the version 1.9 together with some best practices that should be applied when working with containers. Just to cite a few topics touched by the talk: users namespaces, privileged containers, Docker plugins, logging capacities and security aspects like Seccomp and AppArmor.

The last talk of my CloudConf was once again focused on Docker. This time however the main focus was on how to use this technology to create a testing infrastructure able to scale up and down as needed. In this context Emil Stolarsky presented firstly some conceptual principles adopted by Shopify like the distinction between a Scheduler, a tool focused on “how to build”, and a Compute Engine aimed to handle containers and tests distribution. Secondly he described their infrastructural stack, i.e. AWS, and tools like their scheduler BuildKite, their autoscaling technology Scrooge and their images building system Locutus. He then concluded by citing another key principle adopted by Shopify, i.e. “treat your VPS no as pets but as cattle”.

Now I think its my time to conclude this brief summary (there was indeed a lot more! ;)). But before let me say a BIG BIG thank you to all the people that made this event possible. You guys are amazing and I really hope that you can continue to empower an event like the CloudConf! 😀
So…to the next year!

Cheers!

Leave a Reply

Please Login to comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.