ApacheCon EU 2014

The Apache CloudStack Development Analytics Dashboard helps to better understand how the project is evolving and performing. Using information from git, Jira and mailing lists, several charts and numbers provide some insight on relevant parameters of the project. The talk will explain the meaning of the different panels in the dashboard, and will use them and some complementary quantitative data to analyze specific aspects of the inner life of the project. Those will be used to describe the CloudStack community, its activity, and some of their most important development processes.

CloudStack Dashboard: http://projects.bitergia.com/apache-cloudstack/browser/

BaaS(Backends-as-a-Service) and PaaS(Platforms-as-a-Service) are not just for hipsters anymore. 2013 saw many people building their own PaaS like Heroku, Cloudfoundry or BaaS like Parse, Kinvey.

However, many companies and open-source projects only focus on the advantages of building your own *aaS and the advantages of their own solution. Do we really know what works and what should be considered when building such system? Many people do not know whether they should build a PaaS or BaaS, how they should build it, or even if they should build one at all.

Drawing from our experience building both a production BaaS, serving hundreds of millions of requests every month at Korea Telecom, we will explain some of the lessons we learned building large application platforms.

Join us if you are interested in learning more about how to build your own BaaS. The answers may surprise you!

Distributed systems are challenging, but essential whenever running applications on more than a single server. For operators, they pose issues when dealing with machine failures and deploying to multiple machines. For application developers, it often means having to rewrite fault-tolerant features into multiple apps.

This presentation will introduce Apache Mesos and Apache Aurora, two projects that work together to simplify running and deploying distributed systems across large pools of machines by providing a substrate for distributed systems. I'll cover their essential architectures and how they enable fault-tolerance with replicated master nodes, scalability to 10,000's of machines, and resource isolation between tasks. Attendees will come away with an understanding of how these separate software abstractions (resource manager & service scheduler) work individually and together.

Object stores provide a flexible, scalable, and cost-effective abstraction to build applications in both the public and private cloud. Apache jclouds abstracts all major providers, allowing Java source-level compatibility while providing access to vendor-specific interfaces. This talk will cover how to integrate jclouds into applications and how to manage containers and objects. It will also discuss differences between providers, scaling out the number of clients and number of objects, and tradeoffs in application design.