Business Problem: During the requirements gathering, we determined that our client needed to track the progress and performance of the Grid Computing application and be alerted quickly to any problems that developed during a data run.
Solution: We implemented a number of tools to generate metrics and quickly report any that crossed thresholds specified by the client. The metrics included:
- Average performance per PC
- Best performance per PC
- Number of data blocks not completed in the allotted time
- Number of PCs with no data block checked out
Reports were generated automatically on a regular schedule and emailed to designated personnel.
Technical Details: The application data are output in a tab-delimited file, while data about the nodes are maintained in a Microsoft Access database. Using a combination of batch files and awk, the data is processed regularly (via the Microsoft Windows 2000 Server task scheduler) and reports are generated as ANSI text files. These reports are then emailed to designated recipients.