Today, Red Hat focuses on Infrastructure as a Service (IaaS) and Platform as a Service (PaaS). Often, when speaking with organizations about a cloud opportunity I find myself asking questions to find out the appropriate service model for the customer.
“Do you want to just bring your code?”
“Would you like to access the operating system and perform optimizations?”
“How do you feel about kernel semaphores?”
OK, maybe not that last one, but you get the idea. The answers to these questions often help me determine which one of the models, and thereby solutions, to recommend for the situation. With regards to IaaS, Red Hat will soon be providing it’s Cloud and Virtualization Solution – A combination of virtualization and cloud management software that provides the benefits of a cloud computing model with all the underlying virtualization required. For PaaS Red Hat offers OpenShift Enterprise, a solution designed for on-premise or private cloud deployment which automates much of the provisioning and systems management of the application platform stack.
The Synthesized Cloud
Taking a step back, what is the purpose of having separate and distinct cloud computing models? Why couldn’t the models be combined to allow organizations to use elements of each based on their needs? One of the benefits of cloud computing is the ability for organizations to reach the highest level of standardization possible while increasing reuse. If this is the case, then it should be a goal to provide organizations with the ability to utilize not just a hybrid cloud, but a hybrid service model – one in which elements of IaaS could be combined with elements of a PaaS. By realizing a synthesis of IaaS and PaaS service models organizations could leverage the benefits of cloud computing more widely and realize the benefits even in what are often considered legacy, or traditional applications. Cloud Efficiencies Everywhere is, after all, a goal of Red Hat’s Open Hybrid Cloud. I’ll refer to this combining of IaaS and PaaS into a single service model as the synthesized cloud and I believe it is critical to realizing the full potential of cloud computing.
Why not just use PaaS?
Most organizations I have met with are extremely interested in PaaS. They find the increase in developer productivity PaaS can offer very attractive and the idea of “moving the chalk line” up to have developers bringing code instead of hardware descriptions as very exciting. PaaS is great, no doubt about it, but while PaaS can accelerate delivery for Systems of Engagement, it often does not account for systems of record and other core business systems. There is evidence that supports the idea that organizations are shifting from systems of record to system of engagement, but this is not a shift that will happen overnight and in some cases, systems of record will be maintained alongside or complimented with systems of engagement. Beyond systems of record, there are technologies that exist at the infrastructure layer that can be exposed to the platform layer that might not yet be available in a PaaS (think Hadoop, Condor, etc). In time, some of these technologies might be moved into the PaaS layer, but we likely continue to see innovation happening at both the infrastructure and platform services model layers. In short, IaaS finds its fit in both building new applications that require specific understanding of the underlying infrastructure (networks, storage, etc) and as the foundation for hosting a PaaS and consequently will always be important in organizations. For these reasons it’s important to leave our service model open and flexible while simultaneously having a single way to describe and manage both models.
Use the Correct Mix
The ability to use both platform and infrastructure elements is critical to maintaining flexibility and evolving to an optimized IT infrastructure. Red Hat is well positioned to deliver the synthesis of Infrastructure and Platform service models. This has as much to do with the great engineering work and strategic decisions being made by Red Hat engineers as it does the open source model’s propensity to drive modular design.
Some points to consider:
- OpenShift Enterprise, Red Hat’s PaaS, runs on Infrastructure (specifically, Red Hat Enterprise Linux).
- Thousands of other applications run on Red Hat Enterprise Linux (RHEL).
- Application Blueprints provide sustainable, reusable descriptions of any software running on Red Hat Enterprise Linux.
- Red Hat CloudForms can deploy Application Blueprints to a number of underlying resource providers.
Since Application Blueprints can deploy any software running on RHEL and OpenShift Enterprise is software running on RHEL we can deploy a Platform as a Service alongside traditional applications running on RHEL.
Figure 1: Combining PaaS and IaaS
Figure 1 depicts the use of an Application Blueprint to deliver a hybrid service model of IaaS and PaaS. During Design Time, a Developer and System Architect work together to design the Application Blueprint. This involves using CloudForms to define and build all the necessary images that will serve as the foundations for each element in the AppForm (a running Application Blueprint). CloudForms allows the System Architect and Developer to build all these images with the push of a button and tracks all the images at each provider. In this case, a single PaaS Element and two IaaS Elements were described in the Application Blueprint.
The design process also allows the System Architect and Developer to associate hardware profiles to each of the images, and specify how the software that runs on the images should be configured upon launch. Finally, user parameters can be accepted in the Application Blueprint, to allow for customization when the Application Blueprint is launched by it’s intended end user. The result of designing an Application Blueprint is a customizable reusable and portable description of a complete application environment.
Once the Application Blueprint is designed and published to a catalog, users or developers are able to launch the Application Blueprint, the result of which is an AppForm at Run Time. The running AppForm can contain both a PaaS and a mix of IaaS elements and CloudForms will orchestrate the configuration of the two service models together upon launch according to the design of the Application Blueprint.
Imagine an organization has a legacy Human Resources system of record which is a client server model built on Oracle RDBMS. Over time, they’d like to shift this system to a system engagement in order to make it more engaging for their employees. They’d also like to begin providing some data analysis via MapReduce to select individuals in the Human Resources department. In this case, replacing the system of record with a completely new system of engagement is not an option. This may be because of the cost associated with a rewrite or the fact that there are many back end processes that tie into the Oracle RDBMS that cannot be easily changed.
Figure 2: Example Scenario
In this example, the Application Blueprint is designed to include an OpenShift PaaS which delivers a scalable, manged application platform (Tomcat in this case) and both a Oracle RDBMS and Hadoop. Once the Application Blueprint is launched users or developers can access this entire environment and begin working. This goes beyond gaining increased developer efficiency at just the platform layer – it drives many of the efficiencies of PaaS across the Infrastructure as well.
Further Benefits of a Hybrid Service Model
There are many other benefits to this synthesis of PaaS and IaaS service models. One other I’d like to explore is it’s effect on system testing. With a hybrid service model, not only do developers have access to all the qualities of both PaaS and IaaS in a single description that is portable, but the Application Lifecycle Environment framework contained within CloudForms along with it’s ability to automatically provision both PaaS and IaaS can be leveraged to lay the foundation for a governed DevOps model. This provides greater efficiency in testing, accelerating delivery of applications, while allowing for control over important aspects of both the Infrastructure and Platform layers.
Figure 3: Hybrid Service Model leading to Governed DevOps
Figure 3 illustrates how the Hybrid Service Model allows for a governed DevOps model to be implemented. Before the hybrid service model, developers needed to request the required IaaS elements in order to complete a system test. This process is often manual and time consuming. With a hybrid service model in place, upon commit of new code to the source control system, the continuous integration systems contained within the PaaS layer can request a new test environment be created that includes the required IaaS elements for system testing. This greatly reduces the time required to test, and in turn, accelerates application delivery.
How do I get started?
Organizations can begin to prepare for a Hybrid Service Model by ensuring that all decisions made in their IT organization regarding cloud computing adhere to the properties of a truly open cloud. Namely that the technologies the cloud strategy they pursue:
- Is Open Source
- Has a viable, independent community
- Embraces Open Standards
- Allows freedom of Intellectual Properpty
- Allows choice of infrastructure
- Has open APIs
- Enables portability
Red Hat’s Open Hybrid Cloud adheres to the following properties. To learn more about how Red Hat is optimizing IT with it’s Open Hybrid Cloud approach be sure to register for the Optimizing IT Virtual Event which takes place on December 5th, 2012 at 11:00AM EST and December 6th, 2012 at 9:00AM EST.