$[$Analytical Question$]$ Suppose we have a highly connected DCN design as shown in the figure below. Each rack has $10$ hosts and we need high throughput connections on forty flows from hosts i$=1,...,10$ in each of the racks $1,2,3,4$ to hosts i$=1,.\mathrm{.}10$ in each of the racks $5,6,7,$ and $8$ respectively. Each host is connected to the TOR with a $1$Gbps link, and the connections among TORs to Tier$-2$ and Tier$-2$ to Tier$-1$ switches are $5$Gbps$.$ Using this architecture:

$($a$)$ What is the end$-$to$-$end throughput for a host i in rack $1$ to the corresponding host i in rack $5$ in the architecture below? Please explain your answer.

$($b$)$ Can we implement this whole architecture using layer two addressing? Why? Please explain.$[$Analytical Question$]$ Suppose we have a highly connected DCN design as shown in the figure

below. Each rack has $10$ hosts and we need high throughput connections on forty flows from hosts

$i=1,$dots,$10$ in each of the racks $1,2,3,4$ to hosts $i=1,.\mathrm{.}10$ in each of the racks $5,6,7,$ and $8$ respectively.

Each host is connected to the TOR with a $1$Gbps link, and the connections among TORs to Tier$-2$ and

Tier$-2$ to Tier$-1$ switches are $5$ Gbps$.$ Using this architecture:

$($a$)$ What is the end$-$to$-$end throughput for a host $i$ in rack $1$ to the corresponding host $i$ in rack $5$ in

the architecture below? Please explain your answer.

$($b$)$ Can we implement this whole architecture using layer two addressing? Why? Please explain.