Network diagnostics
Understanding Traceroute: How to Read Every Hop
Traceroute maps the path your traffic takes to a destination, one router at a time. This covers how it works and how to read the output.
When a connection is slow or a site won't load, ping tells you whether a host answers. It doesn't tell you where the delay is happening along the way. Traceroute does. It's a diagnostic tool that ships with Windows (as tracert) and macOS and is available on Linux (as traceroute), and it lists every router (or "hop") between your device and a destination, along with how long each one takes to respond.
How traceroute works
Traceroute relies on a field in every IP packet called TTL, or Time to Live. TTL isn't a timer; it's a hop counter. Traceroute sends a packet with TTL set to 1. The first router that receives it decrements the TTL to 0, discards the packet, and sends back an error message identifying itself. Traceroute then sends a packet with TTL set to 2, which survives the first router and gets dropped by the second one, and so on.
By increasing the TTL one hop at a time and recording who reports back at each step, traceroute reconstructs the full path to the destination, along with a round-trip time for each hop.
How to read the output
Each line is one hop: the hop number, the router's hostname or IP, and three round-trip times from three separate probes. Three times per hop, rather than one, make it easier to spot a router that's consistently slow versus one that just had a single delayed packet.
A row of * * * means that hop didn't respond within the timeout window. That's common and usually harmless: many routers are deliberately configured not to answer traceroute probes, even though they pass the actual traffic through without issue. Silence only matters when it continues to the end of the trace, including the final destination. Even then it often means a firewall is dropping the probes, not that the host is down (see below).
What common patterns mean
- A sudden jump in latency at one hop, then it stays high: that hop, or the link right after it, is where the delay is introduced. Often a congested link, or a long distance such as an ocean crossing. A spike at a single hop that is gone by the next one is different: that router was just slow to answer the probe, and the traffic passing through it is fine.
- Timeouts that start at some hop and continue to the end: could mean the destination, or a firewall on the way, is dropping the probes (UDP by default on Linux and macOS, ICMP echo on Windows). Many working sites do this. Check whether the destination itself is reachable another way (an HTTP request, for example) before assuming an outage.
- A route through an unexpected country or network: not necessarily a problem. Internet routing follows business agreements between networks (BGP), not geographic shortest paths, so a "long way round" route is sometimes just how two networks peer with each other.
Limitations worth knowing
Traceroute has real limits. Some routers rate-limit or ignore the ICMP or UDP probes it depends on, which produces gaps that look worse than the actual path. Load-balanced routes can also send each of the three probes per hop over a slightly different physical path, so the three times at one hop aren't always measuring the exact same link. Treat a single traceroute as a snapshot, not a certainty, especially if you're deciding whether to escalate a network issue to an ISP or hosting provider.
The takeaway
Traceroute shows where along the path a delay starts. Combined with ping, a DNS lookup, and an HTTP status check, it turns "the site feels slow" into a specific hop or a specific layer of the stack you can point to. HostChecker has each of these as its own check, and its Host Check runs ping, DNS, and an HTTP status check together in one result, so you see more than one number.