CSC352 Problem of the Day
Homework #4, Problem #2
- Conditions:
- Processing of wiki pages with Hadoop on 6-PC cluster
- Same Mapper and same Reducer program to process two different input folders
| Number of files | Number of wiki pages | Number of categories | Execution Time (seconds) |
|---|---|---|---|
| 589 | 589 | 832 | 388 |
| 1 | 117,617 | 51,120 | 30.7 |
| Ratio=589/1 | Ratio=1/199 | Ratio=1/61.4 | Ratio=12.6/1 |
- Discuss these results
- If you were to add another column to this table, what quantity would you add?
- Identify the parties responsible for this surprising difference
I would add a column showing the number of splits. I think the main culprit is the HDFS, and the fact that a lot of information has to flow through one ethernet switch that is pretty old, and slow...
