Application Centric Computing

Cloud Computing3 minutes readJul 18th, 2012

In May of 2003, I presented a concept called “Application Centric Computing” to a group of Unisys engineers and architects. After that presentation and a lot of revisions, these concepts were presented to many of our largest customers. Their reaction was “Is this really possible?” along with “When will this be possible?”

It’s amazing to see what has happened since then, along with what has not happened.

Here are some of the basic concepts:

1. Application-Centric Computing is the realization of an evolution in computing where IT management can manage how well applications are providing the necessary information and knowledge that is being requested.

  • Each application is managed as an independent entity with the following requirements
    • Service
    • Availability
    • Security
  • Lower level utilities, such as CPUs / Memory / Storage and Networks are self-managed in a manner that is determined by application-focused policies

2. Virtual Application Cells isolate the application from all other applications. It provides a level of management as if the app was running alone on the Operating System, on a secure server.

Concept #1 assumed that we would solve many of the barriers that were impediments to providing utility computing. Most of those barriers have been lifted and large public clouds have shown the way to providing computer resources as a utility. The technology called “auto-scaling” is available for public clouds. This allows policies to be set up that determine when to provision additional VMs in the cloud to meet escalated workload demands.

Regarding Concept #2, a technology similar to “Virtual Application Cells” is available in some grid computing environments, but not readily available for general purpose computing. Given the current state of Unisys thought leadership, we could now reword Item #2 to read:

Hybrid Applications in a Hybrid Enterprise can be isolated from each other, executing as if each application is alone in the Hybrid Enterprise.

But consider another slide from that same deck:

Essential Elements of Application-Centric Computing Remain unresolved

  • What is a “Transaction?”
  • How do you achieve a level of service?
  • How do you achieve a level of availability?
  • How do you enforce security?
    – Access Control
    – Authentication
    – Authorization
  • How do you properly charge for application usage?

This brings up an area that the industry has yet to adopt as standard computing practices. Management is still very infrastructure-focused, but now it’s the virtual infrastruture as opposed to the physical one. Some enterprises understand the value of managing at the application level. They deploy extreme automation at all levels of the infrastructure, including physical, virtual, OS and middleware. Instead of devoting the majority of administration time at the infrastructure level, they focus on the application level. This allows them to work directly with their key stakeholders – business unit leaders and application owners. They measure, monitor, report and control at the application level using tools that are considered part of the Application Performance Management (APM) market.

Applications can be deployed where each application can be specified regarding levels of service and compliance.

The purpose of this blog is to discuss where we were and where we have arrived in the last nine years. And – maybe – where we will be going in the next nine years.