notes

1 compute host fits 28x128MB VMs.

Purpose of experiments:
     1. show that the system works
     2. show the performance of the system - execution time
     3. show how different algorithms perform

What data to report:
     1. The total idle time
     2. OTF
     3. The number of VM migrations
     4. The execution time of different components

What experiments to run:
     1. One set of data, a single run?
     2. Algorithms
     	1. VM placement: BFD
	2. VM selection: min RAM / max CPU
	3. Underload detection: last n average 2
	4. Overload detection: MHOD 0.1-0.2?; last n average 2; LR/LRR 1.0

Output format for metrics - CSV (coma-separated, missing values: NaN):

Algorithm name + parameters | Algorithm name | Parameters | 
Idle time | OTF | VM migrations



Output format for execution time - CSV (coma-separated, missing values: NaN):

Algorithm name + parameters | Algorithm name | Parameters | 
DC iteration | DC VM data collection | DC host data collection |
LM iteration | LM skipped iteration | LM underload | LM overload | LM VM selection |
GM underload | GM skipped underload | GM underload VM placement | GM underload VM migration |
GM overload | GM skipped overload | GM overload VM placement | GM overload VM migration |
GM VM migration