GIGRIB » How it works

How it works

Pingdom GIGRIB users are all part of a large monitoring network. When you install the GIGRIB client, your computer becomes one of the many “test nodes” that report uptime for multiple websites to a central location. In return, all Pingdom GIGRIB users have the right to add ten websites that will be monitored by the network.

For every hour of work your GIGRIB client does for the Pingdom GIGRIB network, you get an additional five hours of uptime checks for the websites you yourself have added. You can accumulate up to 30 days of monitoring this way, ideal if you go on a vacation and will have your computer turned off for days.

Methodology

The main Pingdom GIGRIB servers distribute tasks (websites to monitor) to the GIGRIB clients in the network, while also collecting the resulting uptime statistics from the clients.

• GIGRIB users decide how often their added websites should be checked. Checks can be performed every minute, every 3 minutes, every 15 minutes, etc.
• Each check is made through a HTTP HEAD request and controlled for a valid HTTP response.
• Every HTTP response except 402, 404, and 5XX responses are considered to be valid.
• Every monitored website is checked by several test nodes (clients) in each check interval. If a majority report it as down, the site is considered as down by Pingdom GIGRIB.

How Pingdom GIGRIB determines downtime

If a website is set to be checked every minute, this means that the Pingdom GIGRIB network will check it from approximately five different locations every minute. This redundancy is to counteract any false or incorrect results.

The first thing we do is to sort the results by their time stamps. As an example, here are the sorted check results for one site:

2006-05-16 07:01:28	Up
2006-05-16 07:01:35	Up
2006-05-16 07:01:36	Up
2006-05-16 07:01:41	Up
2006-05-16 07:01:49	Up
2006-05-16 07:01:56	Down
2006-05-16 07:02:05	Up
2006-05-16 07:02:12	Down
2006-05-16 07:02:26	Down
2006-05-16 07:02:42	Down
2006-05-16 07:02:51	Down
2006-05-16 07:03:03	Down
2006-05-16 07:03:17	Up
2006-05-16 07:03:31	Up
2006-05-16 07:03:43	Up
2006-05-16 07:03:56	Up

We look at this data through a “window” which is as “wide” as the check interval of the site (in this case, one minute). This window is moved across the result list as shown below. If more than 60% of the checks in a window indicate downtime, Pingdom GIGRIB considers the website as down.

Note that in the example above, the window we check is always less than or equal to one minute. Downtime is detected from 07:01:56 to 07:03:03 because four consecutive windows in that period had more than 60% “down” results. The first “down” result was at 07:01:56 and the last “down” result was 07:03:03.

Sometimes the number of checks in a window are too small to reliably determine uptime. This is because we do not fully control the GIGRIB clients. If there are less than three checks in one window, we will not take that window into consideration when detecting downtime.