next up previous
Next: Differences Up: Parallelism between PSA and Previous: Parallelism between PSA and

Similarities

Looking at the definitions given in Sections 2 and 3, it is easy to see how the demanded bandwidth Bi is similar to share Fi. The main difference in this parameters is that by definition $\sum_{\Gamma} F_i = 1$, while $\sum_{\Gamma} B_i \leq 1$ (using the constant bandwidth abstraction, a fraction of the CPU bandwidth can be left unallocated).

In [15], the authors show how the real-time tasks' share can be maintained constant at the arrival of a new task ($\Gamma$ varies). This is done by re-arranging the tasks' weights: like for CBA, there is an admission test $\sum_{\tau_i \in \Gamma} \frac{C_i}{T_i}
\leq 1$ (similar to $\sum_{\tau_i \in \Gamma} U_i \leq 1$) and, if a new task is accepted, the weights are recomputed such that for all real-time tasks $F_i = f_i(t) = \frac{C_i}{T_i}$. This is not necessary under the constant bandwidth paradigm, where if a new task is accepted the other tasks' bandwidth have not to be changed.

This difference is due to the fact that Fi is computed based on the executed time, while Bi is computed based on the demanded time. The CBA definition is more natural (doesn't require to change tasks' weights and shares) and permits better resource allocation.

Moreover, with a CBS, the additional parameter Ts (period of the server) can be used to better describe the tasks' temporal behavior. Using Proportional Share schedulers, such as EEVDF or SFQ, it is possible to do a similar thing using a different scheduling quantum for each task, but to the best knowledge of the authors no-one of the existing Proportional Share schedulers actually do it .


next up previous
Next: Differences Up: Parallelism between PSA and Previous: Parallelism between PSA and

1999-02-16