The Mysteries Of Consciousness

The last month or two has been distinctly cloudy, every hosting , telecommunications or commercial company I have spoken to had a single topic on their mind ,cloud computing and what the potential impact will be on their future.The most interesting discussions are with enterprise hosting related companies and shared enterprise data-center operators, where the confusion around “cloud” seem to be most prevalent. Even in this confusion every enterprise co-location, hosting provider and shared data-center owner I spoke to expressed an urgent interest to move into the cloud space, or at least what they perceived to be the cloud space.

The  “xyz as a service” market is currently split into two separate models, with the divide being most clear in the infrastructure as a service (IaaS) market, a market of particular interest to current enterprise hosting providers and shared data-center operators as it has the potential of eroding their current business.

The first model is the enterprise environments as we know today, dominated by large enterprise software providers like Oracle and SAP. The second is the newer market dominated by the Amazon’s of the world. This second market I refer to as Commercial Commodity Computing rather than cloud computing to get away from the cloud confusion. The first trend centers around who leads and who follows in these two markets.

Infrastructure and applications

Trend - Who leads and who follows

Traditional Enterprise

In the enterprise market the infrastructure architecture is dictated by software applications. If you plan a deployment for 100 users you buy a small server. If you plan to deploy for 2000 users you buy a large server. There is no getting around this, nearly all popular enterprise applications have been developed to a deployment model where horizontal scaling is not a first class citizen. In this space the application architecture is fixed and the infrastructure deployment variable.

Commercial Commodity Computing

In the commercial commodity computing space, the application architecture adapts to the infrastructure. As the infrastructure in this space is mostly fixed in dimensions, the application architecture has to adapt to what is available. A good example of how fixed the dimensions are is Amazon’s Elastic Compute Cloud where you have a choice of small, large or extra-large. The application architecture has to fit in this mold to make use of the service. If an application does not fit in one of these molds the only alternative is for the application architecture to be adapted to make use of multiple “servers” or instances to satisfy the requirement. In this space the infrastructure deployment architecture is fixed forcing the application architecture to adapt.

Trend – Buy vs Build

As the commercial commodity computing market is relatively young when compared to the established enterprise software market , developing applications is currently much more prevalent than buying commercial applications. This market is in a similar stage to the enterprise market a few years back when large corporations developed most of their systems in-house, from financial systems to human resource and payroll systems. As the fixed deployment infrastructure building block architecture model becomes more wide spread I expect it to mature in a very similar way to the enterprise model. To a stage where the available commercial applications will be naturally suited for deployment in the commercial commodity computing space.

From the outside any observer would easily think that the IT industry is a string of revolutions, each distinctly new , different and unique in its own sense. Little does an outsider know that it is actually just one long evolution cycle. The confusion comes in when a set of technologies and usage scenarios that have evolved largely in their own domain are put together to provide a bigger benefit and given a marketing name.

One way of identifying this stage in the evolution cycle is when established vendors start renaming or re-branding existing products and offerings to all of a sudden be part of the revolution. Oracle is probably the best example, during the Grid “revolution” tagging multiple Oracle offerings to be Grid offerings without changing much in terms of functionality, just look out for all the “G”’s attached to Oracle product releases :  Oracle Launches Grid Database (eWeek)

I am sure the marketing department now regrets this decision as the Grid concept has yet to move to mainstream adoption outside the technical computing domain where workloads are well suited to Grid processing. A large set of new commercial business applications that suite grid style deployment have just not materialized.

The latest revolution is Cloud Computing, with just about every technology space jumping in on the term not to miss out on the hype going around. Wonder if we will see all the Oracle “G” versions change to “C” to adapt with the times. Oracle 12C in the making ?

Cloud Computing is a culmination of various technologies and compute models thrown into a melting pot of marketing. Software as a service (SaaS), Infrastructure as a service (IaaS), Platform as a service (Paas), Utility computing, Grid computing, Virtualization you name it and it is in there.

This is where the interesting part comes in, these revolutions are marketed as if it is the golden answer to any organizations challenges in deploying an IT infrastructure. For the most part ignoring the fact that most organizations have a large investment in their current infrastructure, likely based on a revolution from just a few short years back. For a moment lets say we can jump on Cloud Computing for our entire infrastructure due to its overwhelming benefit. Surely it is a mature concept and technology. The concept of Cloud Computing under this term is only about a year old, not mature at all , although some of the underlying technologies and concepts have been for years.

Here also lies the answer in how to best approach the revolution in the industry in a more methodical, long term view. As any of these revolutions are based on mostly existing concepts and technologies put together under a new term, assessing the applicability for an organization, infrastructure or application area in terms of more mature offerings is an achievable goal. Much more achievable than chasing a term that is drawn in all directions by vendor marketing campaigns. For example lets have a look at the relative age and inferred maturity of some of the components lumped under Cloud Computing :