The enterprise, government, startups and the system integrators are keenly watching the space of Cloud Computing. While it is a fact that only companies with deep pockets like Microsoft, Google and Amazon can get into the business of providing the Cloud infrastructure, the contribution from other companies is absolutely critical for the Cloud to become viable and real. The biggest concern and risk involved in moving to the cloud is security, privacy and reliability. Most of the customers would want to know what lies beneath the Cloud platform before betting their business on it. Making the Cloud transparent, open and interoperable will enable better adoption by businesses. This is where the Free and Open Source Software (FOSS) will act as a catalyst. The CIOs would want their teams to setup and play with the same stack before actually moving to the Cloud. Knowing that it is the same software that powers their infrastructure at some unknown corner of world will bring a level of comfort to the skeptical decision makers. The Private Cloud Computing offerings are now available from VMWare, Microsoft, IBM, Sun and others. But it is still not clear whether these are the same offerings that power the commercial Public Cloud offered by the Cloud providers. For example, Microsoft categorically mentions that Windows Azure, their Cloud OS is not available as a retail OS that the customers can implement in their data centers. It is also unclear whether Microsoft’s Private Cloud offering based on Windows Server 2008 Hyper-V is the same that is powering its Public Cloud, the Azure platform.
Cloud & Virtualization
Before looking at the critical contribution of FOSS to the Cloud, let’s understand the typical Cloud Computing architecture. At the heart of the Cloud is Virtualization. To bring the elasticity nature to the Cloud, Virtual Machines (VMs) should be dynamically added and removed on-demand. To manage these VMs efficiently, we need a special piece of software that is called the Hypervisor. These Hypervisors can be added to the existing Operating Systems. Some of the modern Server OSs come with the Hypervisor built into them. Ubuntu Server, Red Hat Enterprise Linux ship with KVM and Microsoft’s Windows Server 2008 Hyper-V Edition has a Hypervisor built into it. VMWare, the pioneer in virtualization has some of its Hypervisors like ESX 3.x which are Open Source. The most popular Hypervisor is an Open Source implementation called Xen. Xen is already shipped with the Linux Server editions of SUSE, Debian and few flavors of Red Hat Enterprise Linux. Because the VMs are the workhorses of the Cloud and Virtualization, the OS is just limited to booting up and running the Hypervisor. To avoid the overhead of the OS, Hypervisors are now shipped as standalone layers which do not need a separate OS to boot. Most of these standalone Hypervisiors are wrapped within Embedded Linux. They make the OS completely redundant and can also boot from a USB flash disk. It is hard to imagine this architecture without Linux and OSS forming the core. Though Microsoft had made its Hypervisor, Hyper-V Server free, you still need to invest in costly management software, Microsoft System Center Virtual Machine Manager (SCVMM) to administer, manage and monitor the VMs. The most successful commercial Cloud implementation, Amazon Web Services (AWS) run by Amazon is powered by the OSS Virtualization platform based on Xen.
After looking at the OSS Hypervisors, let’s look at the next layer in the stack, the VMs. VMs are just the virtualized instances of the typical servers that run in the Data Center. These server VMs represent the messaging, database, collaboration & portal, web and application servers. LAMP is undoubtedly the most popular stack that is powering widely used applications on the web including Facebook. When we sign up with Amazon EC2 to run the server instances in the Cloud, there is a huge collection of Amazon Machine Images (AMIs) built on Linux and FOSS and that are available at a nominal price. Remember that you only pay Amazon for the computing power and storage that you consumed and you need not worry about the license fee of the software that you use within these AMIs. For other commercial software, Amazon assumes that you have a valid license to run the software and you are completely liable and accountable for the software licenses. When you got to quickly build a prototype to demonstrate a customized portal app to your customer and you do not have free machines and expertise to setup a LAMP based application, don’t fret! Pickup an AMI from the community images running Ubuntu, Apache, PHP and MySQL along with an Open Source framework of your choice including Drupal, Joomla or WordPress. Just customize these AMIs to your need and store your new AMI on Amazon S3. You are all set to impress your customer.
Apart from the custom Line-of-Business (LOB) applications on LAMP, there are some really powerful frameworks built on OSS. A significant part of the web today runs on Open Source Content Management System (CMS) frameworks like WordPress, Drupal and Joomla. To really exploit the power of on-demand availability of computing power, Apache foundation has released Hadoop. Hadoop is an Open Source framework to process huge datasets leveraging the computing power from dozens of servers. Hadoop enabled The New York Times to convert 4TB of raw TIFF data to an indexed, search-able digital archive of 11 million PDFs in just 24 hours costing them only $240! That is just unimaginable and only demonstrates what Open Source and the Cloud can achieve together.
The other area that is gaining ground is the Private Cloud. Private Cloud promises the benefits of the Cloud Computing while running in the enterprise data centers that are secured behind the firewall. The most popular Private Cloud implementation comes from Eucalyptus Systems. This was started as a research project by the Computer Science Department at the University of California, Santa Barbara before it was distributed through Ubuntu Server by Canonical that promotes Ubuntu and other OSS. The best thing that I personally like about Eucalyptus is its compatibility with Amazon Web Services. It supports the same WSDL as Amazon EC2 and S3. The API and VMs are completely compatible with AWS. I could install Eucalyptus on my Intel i7 Quad-Core, 6GB RAM home machine in a couple of hours to emulate the complete Cloud Controller, Cloud Cluster and Cloud Front-end on the same machine. This was a great experience in understanding the concepts of the Cloud. The other implementation of Private Cloud is from OpenNebula. I am yet to get my hands on this.
When the enterprise and the government want to have a combination of Private Cloud and Public Cloud to form the Hybrid Cloud, these OSS implementations will be really handy. Sensitive customer / patient / citizen data can reside on the Private Cloud thus respecting the privacy and adhering to the local regulations but can leverage the power of Cloud by moving the non-sensitive, compute intensive tasks to a Public Cloud like Amazon Web Services.
I am just scratching the surface of the possible benefits of the marriage between OSS and Cloud. There are many more aspects that I want to share in the coming days.
To catch the demo of OSS on the Cloud, do attend my session on ‘Cloud Computing and Open Source‘ at the Business Technology Summit on November 4th at Indian Institute of Science, Bangalore.
As a Cloud Computing Strategist, Janakiram MSV helps businesses understand and adopt the Cloud Computing paradigm. His core strength is designing and architecting solutions for the Cloud. Janakiram focuses on industry's leading Cloud Computing offerings including Microsoft Windows Azure.
