Diagnosing Network Issues With MTR
Can’t reach your co-located server from some other point on the Internet? Before you contact the data center, run some quick tests with MTR – the problem may be located elsewhere.
MTR stands for Matt’s Trace Route. MTR was written by Matt Kimball back in the 1990’s and was released the public under the GNU General Public License; over the years is has grown and been maintained with contributions by many people.
MTR combines the features of two popular network diagnostic tools – ping and traceroute – into one view. Like traceroute, MTR shows you the route that packets take between two endpoints on the Internet, but it also reports on latency and packet loss occurring at each hop in the data path over time. It does that by pinging each routing point at regular intervals (typically one second) and updating the hop list with data on the ping results. This combined view yields insight into the location of bottlenecks and trouble spots.
The official website for the Windows version of MTR is http://winmtr.net/. There is also a Linux version available at https://www.bitwizard.nl/mtr/. We’ll utilize the Windows version for this quick tutorial:
- Download and install WinMTR.
- Start WinMTR.
- Fill in the name or IP of the host (e.g. google.com).
- Click the Options button to configure ping size, maximum hops and ping interval (the defaults are typically OK). If you want to see host names rather than IPs in the output window, tick the “resolve names” option.
- Push the Start button and wait. One row will appear for each “hop” (routing point) in the path from your computer to the destination host. You will then see updates ripple through the list as additional packets are sent through the path. The columns to the right of the hop IP/name will show the number of packets sent to each hop, best, worst and average response times, and number or percentage of packets dropped.
- You can copy or export the results in text format. Useful if you want to document or report an issue with your provider.
- Click on Clear History to remove the hosts you have previously traced.
Rows in the hop list where latency and/or packet loss numbers are elevated represent potential trouble spots. Keep in mind that some network routers will de-prioritize ICMP packets sent by MTR, potentially yielding higher latency or loss than would be exhibited by the same router with other types of traffic. As a general rule, if a given hop shows high latency and packet loss, and all the hops AFTER that hop also show similar degradation, then you are likely seeing a real problem with that hop. On the other hand, if the hops after the “problem” hop show normal or optimal numbers, then you are likely just seeing the effects of ICMP de-priority.