Importance of visualizing outliers - L1 cache miss rates

Zooming in on L1 cache miss rates as a function of the length of a chain. Note that the miss rates for the three different size packets are shown beside each other, while the two different dataplane implementations are shown with different colors.

The l1 cache miss rates — L1 cache miss rates as a function of the length of the network function chain. The red arrow on the right side of the figure points to some of the outliers that represent packets that had much lower cache miss rates than those covered by the 2^nd quartile (and median) of the box plot. However, these packets underwent exactly the same processing as all of the other packets.

Data provided by Georgios Katsikas, doctoral student at Network Systems Lab (NSL), CoS

Transcript

By zooming in we can see the [miss rate of the] level one cache as a function of the life of the length of the chain for three different packet sizes, and he has spaced the data from the three different packet sizes apart, and he's put them in different colors and used different symbols. And now, we can see here the box plot of data from the two different data plane implementations, and we can see some outliers. And this is really fascinating because we can see these outliers down here, which are nearly an order of magnitude lower cache miss rate.