There are seven stages to a MapReduce job:$-$

Job setup $-$ Load Split $-$ Map $-$ Copy $-$ Merge $-$ Reduce $-$ Write part

Recall Admhal's Law for evaluating the speed$-$up of parallel tasks and your knowledge of which stages can be parallelized among the $7$ stages for the MapReduce Job. Assume job setup, and copy$+$merge take $50$ and $150$ seconds, respectively. Assume with no parallelization, the load split$+$mapper tasks would take $2500$ seconds and the reducers$+$write part would take $1500$ seconds. Given that you have $128$ CPU cores, then based on Amdahl's Law, what would be the achievable speed$-$up in this case?

Round up the result to the nearest integer $($e$.$g$.,4\mathrm{.}6****-->5,4\mathrm{.}3****-->4,$ and $4\mathrm{.}5****-->5).$

Hint: you need to calculate how much of the total time percentage, the non$-$parallelizable $($sequential$)$ stages $($i$.$e$.,*$f$*)$ take and then apply Amdahl's law.