Which hop is causing packet loss?
A traceroute shows every router between you and a destination, with loss and timings for each. It's the right tool for finding where a problem starts, but it's easy to misread. Here's how to read it properly.
If hops and traceroutes are new to you, start with Network basics.
The trap: a router that looks bad but isn't
Routers are built to pass traffic along, and they do that very fast. Replying to a traceroute or a ping aimed at the router itself is a different job, handled separately and given low priority. Many routers also limit how many of those replies they send. So a busy router can show loss or slow replies in a traceroute while forwarding your actual traffic perfectly.
That leads to the single most useful rule:
Real loss starts at a hop and carries on all the way to the destination. Loss that shows up at one hop and then disappears further along is just that router not bothering to answer. Ignore it.
Compare these two traceroutes to the same server:
| Hop | Trace A: loss | Trace B: loss |
|---|---|---|
| 1 · your router | 0% | 0% |
| 2 · your provider | 0% | 0% |
| 3 | 40% | 0% |
| 4 | 0% | 30% |
| 5 | 0% | 32% |
| 6 · destination | 0% | 31% |
Trace A is healthy. Hop 3 looks alarming, but nothing after it loses anything, and the destination gets every reply. Hop 3 is simply slow to answer probes. Trace B has a real problem: loss appears at hop 4 and continues, at about the same level, right to the end. That's where to look.
Finding the hop, step by step
- Trace over time, not once. A single traceroute is a snapshot of a few seconds. Problems come and go, so trace repeatedly across the period when things go wrong, and look at the loss and timings for each hop over many traces.
- Start at the destination. If the final hop has no loss, there's no loss problem on this path, whatever the middle hops say.
- Walk backwards to the first hop where the loss appears and stays all the way to the end. That hop, or the link just before it, is where the trouble begins.
- Work out whose it is. The first hop or two are normally your own network. The next few belong to your internet provider, and later ones to other networks. If loss starts at hop 1, look at your own router, cabling or Wi-Fi first.
- Check latency the same way. Compare each hop with the one before it: how much does it add? A big jump that carries on through every later hop marks a slow or overloaded link. A spike at one hop that later hops don't share is, again, just a slow responder.
Two things to keep in mind
Replies can come back a different way. A traceroute shows the path to the destination, but replies may return by another route. The hop where loss begins is the best place to start looking, not final proof on its own.
Evidence over time is what convinces people. "It was bad earlier" rarely gets a problem fixed. Loss that starts at the same hop, over hours, with times and figures, usually does.
BadHop does this for you. It traces every target automatically (every minute by default), keeps each hop's loss and minimum, average and maximum times across many traces, shows how much latency each hop adds, flags the hop most likely to be at fault, and says in plain English where the loss begins. See how it works, or download it free.