SEARCH
0-9 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
Prev | Current Page 208 | Next

Robert Wrembel and Christian Koncilia

"Data Warehouses and Olap: Concepts, Architectures and Solutions"

Working with blocks of data implies
that the transformation code must be structured in nested loops (an outer ???while
??? and an inner ???while ???), inside which the
transformation macros can take place. The main implementation problem is, however,
how to pass a block of data from one stage to the next and how to exploit the
waiting time correlated to I/O operations.
In Figure 6 we depict the basic scheme we adopted in our infrastructure.
We have two pools of buffers: one for input data and one for output data. A pool of
thread readers fill the input buffers; a pool of processUnits pick the filled buffers,
transform, and write them into the output buffers; and finally another pool of thread
writers downloads the filled buffers into database or flat files.
Aside from implementation details concerning the multibuffer pool, there is the
capability to link one input channel to many output channels, the possibility to
Figure 6. Schema of the transformation hub
thread pool
reader
input buffer pools
output buffer pools
input data
(from LSN, other)
elab processUnit
Loader module
thread pool
writer
0 DBMS,
flat-file, etc.
06 Adzic, Fiore, & Sisto
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of
Idea Group Inc. is prohibited.


Pages:
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220