Fast, dependency-free traceroute in pure C. Minimal syscalls, clear output, significantly faster than standard traceroute.

• IPv4 and IPv6 dual-stack support
• Multiple probe modes: UDP (default), ICMP Echo, TCP SYN
• Output formats: Text (default), JSON, CSV
• Performance metrics: RTT percentiles, jitter, statistics
• Adaptive concurrency: Automatic window sizing based on network latency
• Zero dependencies: Only libc + raw sockets

Fastrace is designed for speed. By using non-blocking sockets, batch I/O, concurrent probing, and optimized syscalls, it completes traces significantly faster than the standard traceroute utility.

fastrace completes the trace ~20x faster while providing richer output including load-balanced path detection.

sudo ./fastrace google.com
fastrace 1.0.0
Tracing route to google.com (142.250.180.174)
Maximum hops: 30, Probes per hop: 3, Protocol: UDP
TTL │ IP Address                              (RTT ms)    Hostname
────┼────────────────────────────────────────────────────────────────
1   │ → 192.168.1.1                           (  6.15 ms)
2   │ * * * (timeout)
3   │ * * * (timeout)
4   │ → 185.89.159.5                          ( 14.77 ms)
5   │ → 37.26.80.129                          ( 14.68 ms)
6   │ → 193.77.91.225                         ( 29.88 ms) bsn-77-91-225.static.siol.net
7   │ → 193.77.107.46                         ( 31.69 ms) bsn-77-107-46.static.siol.net
8   │ * * * (timeout)
9   │ → 142.250.180.174                       ( 37.91 ms) mil04s44-in-f14.1e100.net
      └→ 142.251.235.178                       ( 39.32 ms)
      └→ 192.178.44.134                        ( 38.93 ms)
...
Trace complete in 1523.4 ms. Hops: 13, Responses: 29/39 (74.4%)

traceroute google.com
traceroute to google.com (142.250.180.174), 64 hops max, 40 byte packets
 1  192.168.1.1 (192.168.1.1)  3.876 ms  3.480 ms  3.633 ms
 2  * * *
 3  * * *
 ...
11  mil04s44-in-f14.1e100.net (142.250.180.174)  30.008 ms * *

The easiest way to install on macOS or Linux:

brew install fastrace

Supported platforms:

• macOS (Apple Silicon): Tahoe, Sequoia, Sonoma
• macOS (Intel): Sonoma
• Linux: ARM64, x86_64

make                 # optimized build
sudo make install    # optional, installs to /usr/local/bin

Manual alternative:

gcc -O3 -pthread -lm -o fastrace fastrace.c

sudo ./fastrace [options] <target>

# Standard UDP trace
sudo ./fastrace google.com

# IPv6 trace
sudo ./fastrace -6 ipv6.google.com

# ICMP Echo mode (like ping-based traceroute)
sudo ./fastrace -I 8.8.8.8

# Fast trace with no DNS
sudo ./fastrace -n -m 20 -q 2 cloudflare.com

# JSON output for scripting
sudo ./fastrace --json example.com

# Show performance metrics
sudo ./fastrace --metrics google.com

JSON output (--json):

{
  "target": "google.com",
  "protocol": "UDP",
  "ip_version": 4,
  "probes_sent": 39,
  "responses_received": 29,
  "response_rate": 74.4,
  "hops": [
    {"ttl": 1, "results": [{"ip": "192.168.1.1", "rtt_ms": 6.150}]},
    {"ttl": 2, "results": [{"timeout": true}]},
    ...
  ]
}

CSV output (--csv):

ttl,ip,rtt_ms,hostname
1,192.168.1.1,6.150,router.local
2,*,0.0,
3,185.89.159.5,14.770,
...

Metrics output (--metrics):

── Performance Metrics ──────────────────────────────
  Samples:     29
  RTT min:     6.150 ms
  RTT max:     39.320 ms
  RTT mean:    24.567 ms
  RTT stddev:  10.234 ms
  RTT p50:     25.430 ms
  RTT p95:     38.120 ms
  RTT p99:     39.100 ms
  Jitter:      3.456 ms
─────────────────────────────────────────────────────

If you prefer not to type sudo every time:

macOS (setuid):

sudo chown root:wheel ./fastrace
sudo chmod u+s ./fastrace

Linux (capabilities, recommended):

sudo setcap cap_net_raw+ep ./fastrace

Linux (setuid, alternative):

sudo chown root:root ./fastrace
sudo chmod u+s ./fastrace

Then run directly:

./fastrace google.com

Security note: This allows any local user to run fastrace with root privileges. Use only on personal machines.

# Basic tests (no root required)
make test

# Full test suite (requires root)
sudo make test-full

# Performance benchmark
sudo make benchmark

1. Batch I/O: Uses recvmmsg on Linux for receiving multiple packets in one syscall
2. Concurrent Probing: Sends probes for multiple TTLs simultaneously
3. Adaptive Window: Automatically increases concurrency for low-latency networks
4. Non-blocking I/O: Uses poll() with tight timeouts
5. Optimized Parsing: Fast-path for non-matching ICMP packets
6. Lock-free Counters: Atomic operations for statistics

• Dual socket design: UDP/ICMP/TCP for sending, raw ICMP for receiving
• Background DNS: Hostname resolution runs in a separate thread
• Monotonic timing: Uses CLOCK_MONOTONIC_RAW for precise RTT measurements
• Memory pools: Pre-allocated buffers to avoid malloc in hot paths

*Falls back to standard recvmsg

BSD-2-Clause. See LICENSE.