ISPRS Workshop on Service and Application of Spatial Data Infrastructure, XXXVI (4/W6), Oct.14-16, Hangzhou, China 
In these environments, the hardware, software and sensor 
devices are defined by virtual constructs. The application 
services can likely be designed using virtual constructs which 
are in fine implemented by generic functionalities run by the 
underling infrastructures. 
Because of the duality of the various goals associated with 
emulation and simulation, we have adopted the first approach to 
design and implement virtual computing environments in the 
HEAVEN consortium project. The main benefits are: 
- to provide complex application services deployed on a generic 
infrastructure 
- to be hardware and software independent 
- to be platform independent 
- to be grid infrastructures independent 
- to isolate various topologies of virtual machines from one 
another 
In the latter case, it is possible to deploy and test various 
software and hardware configurations before their production 
use. It is also possible to design specific services for the 
automatic tuning and scaling of infrastructure environments, 
depending on the applications and services being deployed. 
These options make the HEAVEN virtual environments open, 
scalable and generic. Their design allows for legacy software to 
run unchanged, therefore fostering user acceptance. They allow 
new complex applications to be designed involving powerful 
and heterogeneous, distributed resources. They can be tailored 
to each particular applications needs without any consideration 
for resource ownership. They foster new business models where 
resources can be charged on the added application-value and 
not on the resource consumption. They also foster new business 
paradigms where resources can be outsourced to computer 
resources brokers. The applications can therefore be deployed 
and run without the users ever owning the computing resources 
needed by the applications nor the data being processed. 
Figure 3. The HEAVEN infrastructure. 
2.2 Virtual Environments 
By large, complex and production environments dedicated to 
the deployment, monitoring and execution of distributed and 
multidiscipline applications remain to be seen. High- 
performance scientific computing has taken the lead in world 
wide grid computing for two decades. But the price to pay 
however is the expertise level required when deploying and 
running grid middleware, e.g., Globus, UNICORE. 
When compared to the Internet, grids are still in their stone age 
for their ease of use. 
A major concern is therefore the seamlessness of such 
environments. Because new challenging applications are also 
foreseen in such wide-area grid environments, like disaster 
prevention, risk and crisis management, a huge simplification of 
complex computing environments is mandatory. This is what 
we call “breaking the wall”, i.e., breaking the complexity and 
technology barrier hampering the widespread use of state-of- 
the-art technologies for societal and environmental benefits 
(Kotzinos, 2004). 
It is therefore of utmost interest to consider the virtualization 
approaches described above (Section 2.1). 
Clearly, upcoming applications require sophisticated computing 
infrastructures, which are distributed, parallel, multidiscipline, 
heterogeneous and they are used by management teams that are 
not aware of the intricacies of computer technology. 
The deployment environments for these new applications 
require therefore the complex infrastructures handled by the 
simulation approach above, together with the complex 
application environments handled by the emulation approach 
above. 
Figure 4. The HEAVEN approach. 
The future of virtualization, and consequently of virtual 
environments, lies therefore in the conjunction and cross 
fertilization of both the emulation and simulation approaches. 
On the one hand, emulation as defined above (Section 2.1) 
allows building complex environments out of simpler 
infrastructures. They allow for example the design of various 
concurrent dynamic and non-overlapping environments on 
simpler infrastructures. This is fundamental for the design of 
secure and dynamic services tailored to specific and complex 
applications. 
On the other hand, simulation as defined above (Section 2.1) 
emphasizes the use of simplified interfaces to complex systems, 
which is fundamental for the usability and acceptance of 
complex technology infrastructures. In particular, the single 
user view of heterogeneous, distributed and multidiscipline 
computing systems and resources is similar to the required 
interfaces to modem spatial data infrastructures.