Skip to content
bgplookup.io

September 2, 2026

Ping vs Traceroute: Pick the Right Probe

cover ping vs traceroute

Ping answers “does this host answer, and how long did the echo take.” Traceroute answers “which hops answered on the way.” Mixing them up wastes an hour and produces confident, wrong tickets.

Animated traceroute hops versus a single ping
Use ping for health. Use traceroute for location. Use BGP tools when the path itself looks wrong.

What ping is good at

Global ping from many probes tells you whether loss and latency are local to you. A host that replies in 20 ms from Frankfurt and 240 ms from São Paulo is not “down.” A host that drops 40% from every continent is.

Ping uses ICMP echo by default. Many networks rate-limit or drop ICMP. A failed ping with a working HTTPS session is common. Do not declare an outage on ICMP alone if the application answers.

What traceroute is good at

Traceroute increments TTL and waits for time-exceeded messages. Hops that do not send ICMP appear as stars. Stars are not proof of a blackhole. They often mean “this router is quiet.” A blackhole is when probes stop and the destination never answers.

Read the first hop that looks like a different ASN or country. Then look that hop up. If the path jumps to an unexpected origin, open the looking glass.

IPv4 vs IPv6

Run both. Dual-stack hosts fail in one family while the other looks fine. Happy Eyeballs hides this from users and confuses operators. A v6 traceroute that dies at an ISP border with a healthy v4 path is a v6 routing or ICMPv6 filter problem.

Quick chooser

Symptom Start with Then
Is it up from anywhere? Multi-probe ping IP lookup for origin
Slow only from one country Traceroute from that region Looking glass
Packet loss in the middle Traceroute + ping the last good hop Ignore ICMP-only loss on core routers
Wrong website content IP lookup + RPKI Hijack workflow

When you file a ticket, attach ping from two regions and a traceroute, plus the covering prefix and origin ASN. That set ends most “please reboot” loops.

IPv6 probes lie differently

Extension headers, larger minimum MTU, and ICMPv6 being required for path function mean a “broken ping6” can be a real outage even when v4 is fine. Always probe both families when the name has A and AAAA records. A looking glass that has IPv6 routes plus a traceroute6 that dies at hop 2 often points at a CPE firewall that still treats ICMPv6 as optional.

Anycast and “the server moved”

Public DNS and CDNs are anycast. Ping latency that jumps from 8 ms to 80 ms may mean you shifted PoPs, not that a rack melted. Compare origin prefix and looking-glass next hops before you relocate a container. Global ping from many cities is how you see a single sick PoP instead of blaming the application.

FAQ

FAQ

Why is hop 8 lossy but the destination is clean?

Core routers often de-prioritize ICMP generation. Destination success wins.

Should I traceroute UDP, ICMP, or TCP?

TCP to the service port follows user traffic more closely when ICMP is filtered.

Leave a Reply

Your email address will not be published. Required fields are marked *

BGPLOOKUP.IO ACCOUNT

Sign in or create an account

Continue with an account you already use.

Your provider password is never shared with bgplookup.io.