Published . Test run 19 to 20 Oct 2026.
The test
We set up 1,000 test sites to be checked every 30 seconds, from Mumbai, Chennai and Singapore. We added 250 sites checked every minute and 500 checked every 3 minutes. That is about 116 checks a second.
The sites all pointed at our own test server. We could switch it off for all cities or for one city. Every 30 minutes we caused an outage, 24 in all. We measured for 11.5 hours, through the night.
What it used
The main server has 2 cores. It ran at 15-20% load on average. Each city’s checker used 2-4% of one core, at 38-42 checks a second. Two of the checkers ran on small servers with 1 core and 1 GB of memory.
No check was missed, in any hour, city or speed. 99% of checks started within 0.9 seconds of their time, even on the small servers. We saw no memory leak.
The real cost is a burst
Checking itself took little. The busy minute came right after a big outage, when up to 1,750 sites went down at once. Then the database used about 1.2 cores for a minute, to record every outage and send every alert. So the question is not how many sites you check. It is how many go down in the same minute.
Going by the steady load, we think one such main server, with a checker in each of 3 cities, could handle about 5,000 sites every 30 seconds. That is an estimate. We did not measure it.
How fast the alerts came
When the site went down for every city, the usual alert came about 36 seconds later. 95% came within 55 seconds, and the slowest took just under 60. Most of that time is waiting: for the next check, then a second look 15 seconds later, then the cities agreeing. The all clear came within 28 seconds for 95% of the 30-second sites.
Did it cry wolf?
No. In 10 tests only one city lost the site, for 4 minutes. In 3 more, every city lost it for just 8 seconds. None of these sent an alert. In 11 real outages, every alert we expected was sent: 17,585 in all, the exact number. Every site also got its all clear. How cities confirm an outage.
What this means for you
Fast checks do not need a high price. That is why Pro checks every 30 seconds for ₹1,799 a month, incl. GST, and every other paid plan checks every minute. See all plans, or read what a WhatsApp alert looks like.
One test is not a promise. It ran on our test servers, against our own test site. Your site’s speed and network matter too. We will run it again as we grow, and update this page. How we compare.
Questions
- Are 30-second uptime checks expensive to run?
- Not the checks themselves. In our test, 1,750 test sites, 1,000 of them checked every 30 seconds, kept the main server at 15-20% load on average. The cost shows up when many sites go down in the same minute.
- How fast does a 30-second check alert you?
- In our test the usual time from outage to alert was about 36 seconds, and 95% of alerts came within 55 seconds.
- Which TidyMonitor plan checks every 30 seconds?
- Pro, at ₹1,799 a month, incl. GST. Every other paid plan checks every minute.