Seite 39 - Cloud Migration Version 2012 english

39 
Figure 1: TCO‐Reduction with growing size of datacentre (Microsoft) 
Ten years ago, a thousand server datacentre would be considered a large 
scale datacentre. At that time hardware was more expensive, bandwidth 
limited and management less automated. However, technology changes 
over time, as well as the price for technology. During the last decade we 
have seen a dramatic decrease in hardware cost, a decrease in bandwidth 
cost and capability as well as highly virtualised and more automated systems 
management software being created. This has changed the equation on 
what is considered to be the “best economical size” of a datacentre. In 
recent years the IT industry has been able to effectively build datacentres 
with hundreds of thousands of servers and to have a good understanding of 
the economy of scale associated to these new large scale datacentres.   
So, let’s reflect on what the diagram above actually means. As you can see, 
the current combination of the price for hardware, bandwidth cost, labour 
cost, energy cost etc., indicates that a large scale efficient Cloud computing 
provider can reduce 
its own
total cost of ownership (TCO) with up to 80% 
going from 1 000 to 100 000 servers. It’s no wonder these mega datacentres 
are being reported as having the potential to massively reduce the price for 
computing resources for the world.
2.4.2
How does cheaper technology get adopted?  
So, does cheaper technology necessarily mean adoption of that technology? 
If we ignore technical, security or legal considerations for a while, then the