What is Ping (ICMP) monitoring?
Ping sends a small ICMP “echo request” packet to an address and waits for an “echo reply”. If the reply arrives, the machine is reachable on the network, and the measured delay gives an idea of the latency.
It is the lightest and lowest-level test. It is ideal for monitoring network equipment, routers, gateways or servers when you simply want to know whether they are switched on and connected.
Its limit is significant: a machine can answer Ping while the website it hosts is down. Conversely, some firewalls deliberately block ICMP, which gives a false impression of an outage.
What is TCP monitoring?
A TCP test tries to open a connection on a specific port: 443 for HTTPS, 25 or 587 for sending email, 3306 for MySQL, 5432 for PostgreSQL, 22 for SSH, and so on. If the connection is established, a service is indeed listening on that port.
It is the right choice for services that don't speak HTTP: databases, mail servers, internal business services exposed on a port. It detects a stopped service even if the machine still answers Ping.
Its limit: an open port doesn't guarantee that the service is working properly. A database can accept the connection while being overloaded.
What is HTTP monitoring?
The HTTP check behaves like a browser or an API client: it sends a request to a URL and analyses the response. A 200 code means all is well; a 500, 502 or 503 code, or no response at all, signals a problem.
It is the test closest to the real user experience. It detects application failures, web server errors, proxy or load balancer problems, and it measures the perceived response time.
For a website, a shop or an API, it is almost always the main test to set up.
Comparison table
Here is a summary of the three approaches:
- Ping (ICMP) — question asked: is the machine reachable? — ideal for: routers, servers, network equipment — limit: doesn't see application failures.
- TCP — question asked: is the service listening on this port? — ideal for: databases, mail, SSH, non-web services — limit: an open port doesn't prove the service responds well.
- HTTP — question asked: does the site or API respond correctly? — ideal for: websites, shops, APIs — limit: only concerns web services.
Why combine several types of check
The three tests are complementary, and combining them helps you diagnose faster. If Ping, TCP and HTTP all fail, the problem is probably network or hardware. If Ping answers but TCP on port 443 doesn't, the web server is probably stopped. If TCP answers but HTTP returns a 500 error, the application is at fault.
A pragmatic approach is to put an HTTP check on every public site or API, a TCP check on critical services that aren't web, and a Ping on important machines and network equipment. For HTTPS sites, add SSL certificate monitoring, because an expired certificate blocks visitors even if the server is running perfectly.
What LOXUp offers
LOXUp, LOXBOX's monitoring service, supports all three types of check — HTTP, TCP and ICMP Ping — with high-frequency checks, as well as SSL certificate monitoring with alerts before expiry.
Alerts go out by email, SMS (WinSMS in Tunisia) and push notification to all the authorised contacts on your team, and every incident is archived permanently for your analyses.
Key takeaways
Ping for reachability, TCP for a service port, HTTP for the real user experience. For a website, start with HTTP and SSL monitoring; add TCP and Ping for the infrastructure that runs it.
Going further: SSL monitoring · uptime monitoring · all features