Information Technology Reference
In-Depth Information
TABLEĀ 6.1
Initial Test Results
Approach 1: Reduce Total
Execution Time
Approach 2: Reduce Costs
Number of Species
2
4
8
16
32
2
4
8
16
32
Number of VMs
1
2
2
4
8
1
2
2
4
8
VM type
c1.xlarge
c1.medium
Cost per hour (US dollars)
0.66
0.17
Average install time (min)
2.27
2.32
2.13
2.17
2.08
2.52
2.72
2.50
2.25
2.33
Average time per job (min)
10.50
10.85
10.59
10.08
10.10
15.50
15.65
14.30
14.10
14.65
Total execution time (min)
24.77
29.10
49.63
47.12
55.08
38.95
38.58
66.78
66.90
73.90
Used computing hours
1
2
2
4
8
1
2
3
6
10
Processing costs (US dollars)
0.66
1.32
1.32
2.64
5.28
0.17
0.33
0.50
0.99
1.65
Cost per species (US dollars)
0.33
0.33
0.17
0.17
0.17
0.08
0.08
0.06
0.06
0.05
Jobs per VM
2
2
4
4
4
2
2
4
4
4
Finally, it can be seen that the approach in which the total execution time
was reduced has a significantly higher cost than the other one without reduc-
ing the total time in the same proportion. This means that it can be better to
wait a little bit longer for an execution to complete, seeking to improve the
final costs.
6.7 Conclusions and Future Work
Scientific cloud computing is still at an early age. Nevertheless, the academic
community and commercial providers are making important efforts in this
regard. New projects combining public clouds and traditional cluster/grid
approaches will appear over the next few years. Of course, as cloud provid-
ers increase their capabilities to overcome the actual obstacles, new problems
and challenges will appear. This is true not only from a technical perspective
but also from an economical and cultural point of view.
A proposal for a scientific SaaS marketplace has been presented through-
out the chapter. The most important architectural elements were described
in addition to a brief overview of the work done so far in the e-Clouds
project. The solution presented was based on the utilization of the resources
provided by a public IaaS infrastructure, allowing small- and medium-size
groups to access on demand ready-to-use applications while obtaining the
benefit of the scale economies.
The work done so far covers the fundamental aspects of a solution of
such a nature. The design decisions taken so far have aimed toward a func-
tional and simple solution to the requirements mentioned at the beginning
Search WWH ::




Custom Search