Overview
Topic status: We're looking for students to study this topic.
The aim of this project is to implement a memory resident compressed file structure (essentially a RAM disk, in the old terminology). Given a root directory, the program replicates the directory structure in RAM. Files are compressed with ZIP (or similar) and retrieved from memory uncompressed in a transparent manner. The project involves the design and implementation of a RAMDisk class with an overloaded I/O library that supports transparent access to data from disk or from RAM. More advanced features will be studied, using the XML???RAMDisk as an application managed cache, with extended support for XML retrieval. Research questions include the concurrent processing of XML filtering requests (XPath) over large document sets. For instance, in efficiently and concurrently serving requests for filtered XML elements from within files, rather than serving entire XML files. The system will be tested in supporting a high throughput search engine for the Wikipedia with some 50 GB of uncompressed data on HDD, requiring about 6GB of RAM when compressed. The specific objective is to perform concurrent processing of 1000 Wikipedia queries utilising multithreaded execution on a modern multicore multiprocessor PC.
Pre requisite - very good programming skills in Java and/or C/C++
- Study level
- Honours
- Supervisors
- QUT
- Organisational unit
Science and Engineering Faculty
- Research area
- Contact
-
Please contact the supervisor.
Associate Professor Shlomo Geva
