Are there universal scripts/libraries for the analysis and definition of non-standard situations in the system?
The issue of discharge "prompt where to dig?".
1. I'm looking for something that will help you to track some system events. For example, Google Analytics is service Alerts, helps you increase the specified (percentage) threshold of site visitors to send notifications, in fact, this very excess. There is also a lot of monitoring systems of the servers that send the notification if the server is a virtual machine or something has died. I called this logic "comparison with a fixed reference value".
2. Consider a slightly more complicated thing. Imagine that we need to track is not the achievement of a certain threshold, and the deviations from steady state. From my point of view it is feasible in the following way: something (assuming it is a script) receives the input of certain values and compares them with the accumulated statistics for the previous time interval. If the current value exceeds the average value for the interval, then the event happens and the system owner is made aware of this. This example is in the "comparison with floating reference value".
3. Now even more complex behavior. If the first two options we worked with well-known events (and conventionally, we can program the logic on a simple if'Ah), now I want based on the input (and historical) data system to create events: to analyze input, and identify the recurring patterns and to create events, which can already be applied are described in the first two paragraphs of logic.
Actually, the question is: are there such algorithms/libraries/scripts?
To clarify: Pro data-mining, bi, clustering algorithms and everything associated with it I read. I'm more interested in specific products and experience.
There recently was a good article on monitoring VoIP telephony, and the plot, try to search in the appropriate blog/tags — blijaishie a couple of months.
here's even a link to the article I found habrahabr.ru/post/132400/
Herminia.Kuhic answered on October 8th 19 at 02:09
Such systems are tied to technology. But the implementation often falls on Your shoulders. Algorithms can be sought, such as exception handling — if the system provides, then all is well, and if not, then the editor in your hands! An example of such a technology D-BUS. More words to describe the product/system You want to monitor?