Advances, Systems and Applications
Table 4 Locality levels achieved by each scheduler based on different job size
From: CLQLMRS: improving cache locality in MapReduce job scheduling using Q-learning
Job size | FIFO | Delay | ACL | CLQLMRS | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
CL | DL | RL | CL | DL | RL | CL | DL | RL | CL | DL | RL | |
3 Maps | 0% | 2% | 50% | 12% | 75% | 96% | 21% | 81% | 99% | 28% | 85% | 99% |
10 Maps | 7% | 37% | 95% | 19% | 99% | 100% | 29% | 99% | 100% | 36% | 99% | 100% |
50 Maps | 12% | 61% | 98% | 24% | 95% | 99% | 38% | 98% | 100% | 43% | 99% | 100% |
100 Maps | 25% | 78% | 99% | 35% | 91% | 99% | 46% | 94% | 100% | 55% | 98% | 100% |