# Archipelago Monitor Statistics

Archipelago (Ark): CAIDA's active measurement infrastructure serving the network research community since 2007.
Statistical information for the topology traces taken by this individual Ark monitor is displayed below. See the main statistics page for the full list of monitors

Prev monitor: cld4-us
Prev in US: cld4-us
Prev in North America: cld4-us

Next monitor: cld6-us
Next in US: cld6-us
Next in North America: cld6-us

# cld5-us

Spectrum
San Diego, CA, US (66)

### CCDF of destination RTTs

• percentile 10th 25th 50th 75th 90th
RTT (ms) 70.046 99.100 170.852 207.475 265.480
Use the following link to download the data used to render this graph in ASCII, comma-separated values format here: (CSV output)

### Description

This graph shows the complementary cumulative distribution function (CCDF) of round-trip times (RTTs) to the destination host.

### Motivation

By showing the distribution of RTT values to all responding destinations, we can get a sense of how varied the speeds are for connecting to different points in the Internet.

### Background

The complementary cumulative distribution function shows the fraction of collected data points that are greater than a given value. This is backwards from how percentiles are given, as those show the percentage lower than a given value. On this graph, you would find the 80th percentile at the 0.2 Y value. The round trip time of a probe is the time (in milliseconds) that it takes for a packet to be sent from an Ark monitor to a destination and for that destination's response to be received by the monitor. Therefore, no RTT values are recorded when a probe does not reach a destination.

### Analysis

When the CCDF graph has a nearly vertical dropoff point, that indicates that RTT values fall within a narrow range. This tends to mean that a bottleneck exists within the monitor's connectivity that dominates over individual destination path variation. A more gradual curve, on the other hand, indicates greater variability in the response times of destinations, which tends to scale directly with the path length distribution.