Cloud is the new mainframe

Cloud computing, from a business and management perspective, has a great deal in common with mainframe computing.

Mainframes are powerful, expensive and centralized pieces of computing equipment. This is in line with their role as infrastructure for mission-critical applications. For these types of applications, mainframes can be fairly efficient, even though they tend to need large teams of support specialists. Mainframes are often associated with ‘legacy’ applications, and technology lock-in of the organization to those applications and associated platforms and vendors.

Cloud computing is a new style of computing, and in particular a style of computer deployment that differs from racking and stacking lots of x86 servers in company data centers. From a business and management perspective this difference signifies a return of some of the characteristics of mainframe computing, in terms of power, expense, and centralization of operations and expertise.

Cloud computing is undeniably powerful and capital intensive. Although exact numbers are hard to get, the largest cloud providers have spent billions of dollars on their infrastructure. (Microsoft Dublin, Google).

To emphasize this point, Werner Vogels (CTO of Amazon) has the lid of his laptop laser-engraved with the text “My other computer is Amazon EC2”.

Cloud computing is also fairly centralized in a few data centers around the world. Most importantly, these centers are remote from the perspective of the users, which has its own technical implications. A lot of specialists and competencies are needed to use cloud computing well. For example, it is not always trivial to optimally harness the cloud’s inherent scalability. Depending on the exact cloud model chosen, these specialists would work for the cloud providers or for their customers developing cloud based solutions.

Let us have a look at some of the competencies and capabilities that will become in demand again. My mantra for that is: “Those who cannot remember the past are condemned to repeat it.” as poet and philosopher George Santayana said in 1905.

The mainframe is a computing infrastructure shared by a number of applications. The average x86 server silo farm is not. Cloud computing, in any shape or form, is a shared infrastructure. This (re)introduces the necessity of managing capacity, cost and performance. On the mainframe, capacity and performance management is required to resolve resource conflicts between applications, as well as to properly plan for (or avoid) expensive upgrades. As an example, reengineering the date conversion routine has saved many companies more than a million dollars worth of upgrades. The server silos model reduces the apparent need for stringent capacity planning, though at considerable capital expense. While cloud computing has radically different flexibility and capital expense models, the manifest need for capacity and performance planning resurfaces. This time however, it is not just about managing competition for resources, it is directly about managing operational expense.

Cloud computing also reemphasizes the need for transaction monitoring, although that now has names such as “end user performance monitoring” or “application performance monitoring”. In fact, this has become even more important and complicated, as more parties are now involved. The end-user is no longer solely on company premises, but could be an e-commerce customer at home, or a business partner working remotely.

There is a vast body of knowledge and a great number of practicing professionals working in this space. The prime example is the Computer Measurement Group, which has been active since 1974. While its roots are undeniably in mainframe computing, the yearly international conference also has a long tradition in addressing capacity and performance of computer networks, client server architectures and, in recent years, cloud computing. CMG could very well rebrand itself as “Cloud Measurement Group”. Even so, within and outside this group, this expert knowledge, the skills of people, and the measurement and management toolsets all need updating.

“I think there is a world market for maybe five computers.” — Thomas Watson, chairman of IBM is alleged to have said around 1943. In a twisted way, this may become somewhat true again. In the future there might well be five computers worldwide, their names could be Google, Amazon, Azure, Salesforce and Facebook.

Clearly things have changed a lot in the last fifty years and will increasingly continue to do so in the coming years. For IT people willing to embrace change, there is ample opportunity, but reskilling is necessary. Fortunately, a lot of traditional IT skills can be upgraded to the cloud world.

My own professional path is focused on this. I am training people worldwide on cloud computing and virtualization, and contribute actively to CMG. During the coming annual CMG conference (Dec 2011) I will conduct a workshop on cloud computing for capacity planners. To further support people in this field I am also developing www.clubcloudcomputing.com. Feel free to join me in this journey.