Track: Cloud Computing

Theme: CODE

On: Oct 31, 2014, from 09:00 to 12:20

Track leader(s): Jonathan Le Lous (Infrastructure Solutions Director / Cloud leader, Savoir-faire Linux)

You think the cloud revolution is ending ? Only one way to reach on ? You're wrong, that's only the first step. In this sessions we will discuss about DataBase scaling, IaaS solutions and new community initiatives. Each day the cloud is more opening. Don't miss to update your knowledge and come to Open World Forum !

Talks

09:00 - Fedora Atomic

Duration: 45 minutes

Speakers: Haïkel Guémar (Contributor, Fedora Project)

Docker has revolutionize the way we ship and deploy applications, but it's only a building block. Fedora Atomic is a secure and reliable way to host Docker containers and managing your apps and OS as atomic units (hence the name). It is structured around technologies like: Docker (obviously), SELinux, rpm-ostree, geard

09:55 - Cloud Coalescence: The Collision of Virtualization and Containers

Duration: 35 minutes

Speakers: Brian Proffitt (Community Liaison, oVirt and Project Atomic, Red Hat)

Until recently, cloud computing has been about virtualization. Virtual machines are at the heart of the cloud, bringing cost savings and flexibility to organizations on a massive scale. But there's a new weather system blowing in: containers. Instead of emulating an entire operating system, containers can run just enough code to efficiently run an application. Containers and virtualization can be construed as competing technologies in the cloud, but in reality they are different parts of the same cloud ecosystem. From oVirt to RDO to Project Atomic, open source projects are leading this new innovation that will transform cloud computing as we know it.

This talk will detail the new container technologies within the cloud, and put together a map of how virtual data center management, containers, and cloud computing will all work together.

11:00 - How to build an elastic MySQL database

Duration: 35 minutes

Speakers: Serge Frezefond (Cloud Solutions Architect, MariaDB)

How to build an elastic MySQL database is a main challenge for large scale data centric infrastructures.

A complex problem facing many successful projects is how to handle a sharded database environment. We will take a look at the possible solution to address this difficult problem. Building an elastic database is also a main issue in for cloud providers wanting to offer an elastic database as a service.

In most data centric application the data size growth and the number of requests growth cannot able to handle by a single server. A common technique for scaling is to shard the database thereby reducing the write load and data size for each server.

We will examine various sharding solutions currently used by big web properties (Google, Facebook, Tumblr ...).

We will also go through two new and more generic approaches : MySQL Fabric an MaxScale.

MySQL Fabric is a framework to manage a farm of sharded and highly available servers. MySQL Fabric is associated with sharding aware connectors (PHP, Python, Java …) that provide an enhanced API for working with MySQL Fabric sharded databases.

MaxScale is a smart proxy which allows to transparently access a set of servers. MaxScale offers advanced capabilities like filtering, connection load balancing, read/write splitting. MaxScale is fully transparent for the client application and does not requires a specific connector.