From “it's slow” to “AS64500 is dropping packets since 14:05”.
Synth Monitor diagnoses every test continuously. Here is the method, so you can trust the answer, and check it.
Measure from many places
Agents at your sites, on laptops, in your cloud accounts and in our regions run the same test. Each run records the result and, for network tests, the hop-by-hop path over several flows.
Decide what is really degraded
An agent counts as degraded when its latest run fails, or loss or latency stays above its own normal for two runs in a row. One slow run is not an incident.
Find where it starts
On each degraded path: the first hop where real loss begins and continues downstream, where traffic stops, or where latency jumps against the last healthy path. Rate-limited routers are ignored.
Name the network
Every hop is mapped to its AS and a role: your private network, your access ISP, a transit provider, the destination. The culprit is reported by role and name.
Compare across agents
If the affected agents share the culprit network it is a provider problem; one site alone is local; all agents failing at the target is the destination; resolution errors are DNS.
Line it up in time
Every path is kept. Route changes and BGP bursts on the destination's prefix just before the problem began are added to the diagnosis, and you can slide back to any moment.
One sentence, then the evidence.
The diagnosis leads with what is wrong, where and whose network, then lists each agent's problem and where it starts. It also says what is fine, because “your network and the destination are fine” is often the most useful line in an incident.
| Agent | Problem | Where |
|---|---|---|
| Frankfurt-1 | 14% loss | hop 4 · AS64500 |
| London-2 | 13% loss | hop 5 · AS64500 |
| Paris-Office | 12% loss | hop 6 · AS64500 |
| Singapore-1 | Healthy | via AS64511 |
Why most traceroutes lie, and how we don't.
Routers answer traceroute probes from their control plane, at low priority and rate-limited. A hop showing 40% loss while every hop after it shows 0% is not dropping your traffic. Synth Monitor only counts loss that persists to the next hops and the target, and judges the target by the test's own packets.
Several flows per trace reveal load-balanced paths: a hop that answered on two flows but not on two others is a branch, not loss.
An evidence report they will act on.
The report holds the verdict, a UTC timeline, the source networks and the full hop-by-hop paths with networks, ready to print, to share as a public link, or to turn into an email with one click.
Target: api.example.com (192.0.2.10) tcp/443
Problem since: 2026-10-21 14:05 UTC (ongoing)
VERDICT: packet loss begins in AS64500 Transit-Example
first lossy hop: 203.0.113.14 (hop 4), 14% loss
loss persists to the destination: real, not ICMP rate-limiting
3 of 4 source networks affected; AS64511 path healthy
TIMELINE (UTC)
13:59 route change: AS64496 → AS64500 → AS64510
14:05 loss starts (Frankfurt-1, London-2, Paris-Office)
See whose network it is, on your own traffic
Free during the beta. Install an agent in one command and your first test is live in minutes.