Menu Content/Inhalt
Home arrow About Us arrow Compucon Behaviours arrow 2013 Computing Technology Seascape 80/20

2013 Computing Technology Seascape 80/20 Print
July 2013

This note is based on observation and is not theoretical, academic or experimental.  It attempts to look for patterns of change and to categorize the current situation in June 2013 for the purpose of providing clarity.  This note is debatable and must be read with caution and shall not be relied on for making decisions without a direct discussion with the Compucon system architect.

The last 120 years (not 12 years) has given birth to 3 eras of computing technology- vacuum tube era, semiconductor era and the current digital IP era.  Each progressively expanded computing from central organisation based to personal desktops to mobile collaboration over the Internet.   The size of computing hardware has shrunk over 4 periods- the periods are mainframe, mini, micro, and mobile (SOC) where SOC stands for System-on-chip.   Market behaviour has also changed over each period.  Initially people chased for performance since it was scarce; then for price as computing became more commonplace; and then for power (electrical efficiency) when computing was needed all the time.  None of them was disruptive on its own.  Good enough has been responsible for the emergence of lower performance, lower power, and lower priced technology for mainstream adoption!

We divide computing into two categories, mainstream and niche, with a population ratio of 80/20 respectively for convenience. 

o Mainstream computing users are those that use thin clients and devices that are mobile.  They use cloud computing as well.  At the backend cloud computing providers are large global organisations; their customer base and infrastructure spread across national boundaries.  To handle the global population, these providers adopt commodity hardware and deploy software as an abstraction layer.

o We refer niche computing users to those who are not mainstream users.  Supercomputing data centres are one example group.  SKA is another typical example. Their approach to computing is rather different.

o Supercomputing has used the best and a lot of top of the line commercial off the shelf (COTS) hardware.  Typical hardware suppliers are as IBM, Intel, Nvidia and AMD.  Computational performance is the main issue for them. 

o SKA Central Signal Processing is for real time streaming and processing.  It is designed for specific functions and while it also involves a large quantity of hardware, COTS hardware may not be the most appropriate solution and a good balance between hardware cost and programming effort required must be achieved. 

Digital is binary (zeros and ones).  The real world is not.  There is a continuum connecting the mainstream and niche computing categories described above.