1: \begin{proof}

2: For $N-K < K$, we show that not utilizing the systematic node can only add load to the system, and thus reduce its service capacity region. Suppose $\lambda_i < \mu$ for some $i$, that is all requests for $f_i$ can be served by the systematic node. Instead, suppose we serve $\lambda_i - \epsilon$ rate using the systematic node $i$, and send the remaining $\epsilon$ portion to $K$ other servers, and decode file $f_i$ from the coded versions. As a result we are reducing the load on the systematic node by $\epsilon$, and instead adding $\epsilon$ load to $K$ other servers. If $N-K < K$, at least one of these $K$ servers is also a systematic node, which stores file $f_j$. Thus, the maximum rate of requests for file $f_j$ that can be served by its systematic node reduces by $\epsilon$.

3: 

4: For $N-K > K$, we showed in \Cref{thm:waterfilling_opt} that the water-filling algorithm, which first sends requests to the systematic node is optimal. Thus, there is no loss of optimality in sending requests to the systematic node until it is saturated.

5: 

6: %\TODO{Make this lemma and proof rigorous}

7: \end{proof}

8: