What is a TCP ping?
A TCP ping measures whether a host is reachable and how fast it responds by opening a TCP connection rather than sending an ICMP echo. Because ICMP is unavailable behind most hosting and CDNs, IPeek measures reachability and latency with a TCP connection to ports 443 and 80, reporting the minimum, average and maximum round-trip time over several attempts.
How TCP ping differs from ICMP ping
The classic ping command sends an ICMP echo request and times the reply. That works on a local network but fails across much of the public internet, because firewalls, CDNs and cloud providers routinely drop or deprioritize ICMP. A TCP ping instead opens a real connection to a service port and times how long the handshake takes. Since web servers must accept connections on 443 and 80 to function, this reliably reaches hosts that never answer ICMP, and it measures the latency of the path that actually carries your traffic.
How to read your latency results
IPeek reports minimum, average and maximum round-trip time across several attempts. The minimum is the best-case latency of the path; the average is what you can typically expect. The gap between minimum and maximum is jitter: a small gap means a stable connection, while a large spread signals congestion or an overloaded host. As a rough guide, under 50 ms is excellent, 50 to 150 ms is normal across regions, and consistently high or wildly varying times point to a network or server problem worth investigating.
When a TCP ping is the right tool
Reach for a TCP ping when a plain ping reports 100% packet loss but the site clearly loads, which simply means ICMP is filtered. Use it to confirm a server is accepting connections on its web ports, to compare latency to a host from your location, or to spot intermittent slowness before it becomes an outage. It is also the honest measurement when you care about real user experience, because it times the same TCP path that browsers and APIs use, not a separate ICMP channel.
Why ICMP is often blocked
Operators block or rate-limit ICMP to reduce attack surface and noise: ICMP floods were a classic denial-of-service vector, and echo replies leak information useful for reconnaissance. CDNs and load balancers also terminate connections at the edge, so an ICMP ping would measure the nearest edge node rather than the origin. Because of this, a failed ICMP ping says nothing about whether a site is up. A TCP ping to the ports the service actually listens on gives a reliable yes-or-no on reachability.