next up previous contents
Next: Results Up: Deterministic CPISync Previous: System Model   Contents

Metrics and Measurements

We consider the time required for a synchronization to complete as the most important metric in our experiments. There are two factors that jointly contribute to the total synchronization time - the data transfer time and the computation time to synchronize. For Slowsync the dominant component of the latency is the data transfer time, whereas for CPISync the computation time generally dominates. Our experiments compare the latencies of CPISync and Slowsync in various scenarios. The synchronization latency is measured from the time at which the PDA beings to send its data to the PC until the time at which the PC determines the differences between the databases. On the down link from the PC to the PDA, both Slowsync and CPISync will transmit the same updates.
Figure 3.5: Comparison between (a) Scalability with the size of the data sets (b) the time complexities of CPISync and Slowsync for increasing differences.
\begin{figure}
\centering
\subfigure[]
{\epsfig{file=fig5of3.eps,width=7c...
...ure[]
{\epsfig{file=fig6of3.eps,width=7cm}} \vspace*{-0.25in}
\end{figure}

next up previous contents
Next: Results Up: Deterministic CPISync Previous: System Model   Contents
Sachin Kumar Agarwal 2002-07-12