Information Technology Reference
In-Depth Information
Table 14.1:
Technical profile of IT tasks performed in companies (Messerschmidt 2009)
Financial services industry
Remaining industries
Decomposition into sub-processes possible
64%
61%
% of tasks that can benefit from increased
computing power of one server
35%
31%
% of tasks that can benefit from an increased
number of servers
31%
29%
Decomposition into sub-processes
not
possible
36%
39%
% of tasks that can benefit from increased
computing power of one server
19%
24%
% of tasks that can benefit from an increased
number of servers
18%
23%
Across all industries, in more than 60% of all cases some of the tasks can be decom-
posed into sub-processes. For the financial services industry, in companies where
some tasks are decomposable, 35% of tasks could benefit from greater computing
performance of a single resource and 31% from a greater number of resources allo-
cated to the task. Both can be achieved via a Grid. Sub-processes could be spread
over a great number of client PCs that individually have low performance, but
collectively, as a Grid, provide significant compute performance that can drastically
reduce the computation time of a task. Table 14.1 also shows that in companies
where tasks are not decomposable, the number of tasks that could benefit from a
Grid is relatively lower.
The majority of IT computing tasks in the respondent companies - 75% - are
database queries and 81% of computing tasks are requests for data storage, while
only 36% are dedicated to Web applications. Only 32% of the tasks are dedicated to
high performance processing tasks and 23% are dedicated to redundant calculations
in order to increase the accuracy.
This type of tasks are different compared to HPC tasks, which are typically the
main target tasks of Grid Computing and illustrate the different needs in industry
compared to eScience. The 36% of Web Application indicate a substantial potential
for Cloud Computing as Web applications are suitable to be run on Clouds.
The survey revealed also that a substantial number of companies (58%) needed
up to 12 months for building the Grid until the Grid was operational; after two
years a cumulative 92% of companies were able to use the Grid. In only 10% of
the companies the implementation of the Grid solution took longer than three years
(Messerschmidt 2009). This confirms that the introduction of Grid Computing is a
complex and long-lasting process.
The survey involved also questions related to the technical and non-technical
obstacles for introducing Grid Computing. The 81% of the companies participating
in the survey that are not using Grid Computing mentioned the following major
technical obstacles (Messerschmidt 2009):
