Seite 41 - Cloud Migration Version 2012 english

41 
Energy Costs 
When the cost for hardware decreases, the proportion of the energy cost 
for running the servers increases. It may add up to 20% of the total cost and 
for a low‐efficiency datacentre; the three‐year spending on power can be 
larger than the three‐year server hardware spending. The power usage 
effectiveness (PUE) for large datacentre tends to be much lower than for a 
small datacentres resulting in a reduced cost. Also, a super datacentre may 
consume up to 100 MW of energy. Buying this much energy and placing the 
datacentre in locations where energy is clean and cheap normally means 
getting much better energy prices than those available to the average 
consumer.  
Labour Costs 
The high levels of systems management automation undertaken in the big 
datacentres may reduce labour cost per server dramatically.  In a typical in‐
house datacentre there is one administrator for each 100‐200 servers. In big 
Cloud datacentres there is typically one administrator taking care of several 
thousands of servers.  
Hardware and infrastructure 
A company buying 1 000 servers does not have as much buying power as a 
large datacentre buying 100 000 servers. If you install this many servers, it 
may make sense to tailor the hardware for the environment to further 
decrease the hardware cost. Removing things like the cover, sharing power 
transformers or using CPU’s that demands less cooling.  
All in all, these factors make it much more cost effective for a Cloud provider 
to build a big datacentre than a small one.  
2.4.4
Why is it cheaper for the consumer? 
Normally, an ordinary consumer does not have the same possibilities to 
reduce its own datacentre costs to the same low cost as a large Cloud 
provider. It is very difficult to estimate the exact savings the Cloud may bring 
to a consumer as it depends on many factors, some companies’ benefit 
better from Cloud computing than others. 
For early adopters, the usage of Cloud‐based services are often companies 
that have a large variability in its business. It can be companies in retail (who 
may sell 3x their usual target in December) or the tax authorities (who may 
have 10x the load during tax season). These organisations may have been