Archive for November, 2015

Azure and Open Source

November 2, 2015

It may come as a surprise to some that Microsoft has been working with open sources solutions like Linux for a number of years now. Azure is a good example of how Microsoft incorporates open source into it’s offerings.

  • Microsoft showcases the Azure Cloud Switch –  Azure Cloud Switch (ACS) is built on the Switch Abstraction Interface (SAI) which is a Linux based open source solution that allows Microsoft data centers to efficiently manage the many Azure Applications.

Azure Cloud Switch

  • Microsoft Extends Commitment to Open Compute Project – Want to know the exact hardware that  Azure and Facebook are running in their Data Centers? Take a look here:

azure hardware


AzureCon is a great way to get up to speed with all the things that are going on with Azure. Below are some of my favorite sessions:

  • Fundamentals of Revolution R Enterprise – I have been using R for a while now and while it has a great collection of packages, it also has some serious limitations. To begin with R is only in-memory (no disk storage available during data processing) and it is single threaded (there are work arounds but they take a lot of work to get right). R Enterprise is multi-threaded and has disk representation (.xdf file). It is interesting to note that the R Enterprise data file is columnar.
  • Data Warehousing In Cortana Analytics – SQL Data Warehouse (SQL Data Warehouse documentation) is the Azure version of PDW (Parallel Data Warehouse) and both have the same code base (APS Documentation ).
  • Deep Neural Networks – Neural Networks modules in Azure Machine Learning have a script input for determining the topology. The scripting language is called Net#. With Net# you can control the number of layers and how they connected including transformations like convolutions. This means that these settings can easily be shared across different models without having to build each model from scratch.
  •  Data Science with Microsoft Azure and R – I recently completed a the six-week class (edX Data Science and Machine Learning Essentials) which is a great introduction to  Azure Machine Learning for Data Scientists.