This page has been robot translated, sorry for typos if any. Original content here.

We organize web server health monitoring

Web servers on the Internet are like patients in a hospital ward, for whom constant care is needed. I almost didn’t understand - also my favorite server doesn’t ping in any way, forcing the network ruler to get excited about creative ideas about the upcoming clashes with equipment, provider, clients, management ... what to do? Do you also want to sit on duty all night long on the computer? Or conduct manual monitoring from a remote machine? No, this is not our method. We will try to find something automated, affordable and convenient. The Google Network Search gurus, by swallowing our “website monitoring” request, gave a great deal of web-browsing links to various resources, offering their services to the county audit and also reviewing the performance of web servers. Monitoring systems (in the Basurman language, they are often called uptime-systems) are quite a few - also paid, also gratuitous. We will not consider the latter in any way due to their limited capabilities and the lack of job security of the service they provide.

So, it is decided - we choose a commercial monitoring system. What else should you care about? First, look at the presence of several monitoring points (monitoring checkpoints), preferably geographically distributed across different servers. After all, you see, it is foolish to delay the detailed and reliable statistics collected only from one computer (polling point). Secondly, inquire about the list of services provided as well as services. This list can be very extensive - from testing HTTP, POP3, IMAP4, FTP and other protocols, up to monitoring DNS also other services. Look at the capacity as well as the quality of the statistics provided - it will help to investigate the chronicle of the “ups and downs” of your servers throughout the entire period of their monitoring. Additionally, in the end, thirdly, you must carefully approach the task of parsing the convenience of using the system. By polling your servers, the system will generate a walk of a variety of reports, with which you will have to work, also to draw any conclusions.

Today we will try, based on the above tips, to take at random some commercial service for testing.

For example, WatchMouse.

The first good news, which at once rushes into the eye - this is a gratuitous 30-day period of using a full-fledged monitoring system. The visionary developers of WatchMouse have acted quite cleverly, because the capabilities of their system bribe users, but it is rather difficult to refuse a tasty one. What blah blah allows this "observant mouse"? We list the points.

• Analyze servers using multiple monitoring points located throughout the universe. The spot of the poll to eat even in Rio de Janeiro - Ostap Bender would be glad.
• Detailed text and graphical statistics, including the ability to export data to Excel.
• Powerful event filtering.
• Availability of various online networking tools (Ping, Traceroute, DNS Query and others).
• Notification of events via e-mail, ICQ, SMS as well as access to statistics via WAP, using mobile phone capabilities.
Imagine what amenities provide modern IT-technology? It is terrible to think that only one SMS with information about the "fallen" web server can destroy the glamorous illusions of the system ruler in a flash and sit in some place in a cozy cafe and drink a good cool beer ...

Well, what else is allowed to add? In my opinion, we told about all the main points also the intricacies of choosing systems for monitoring. We only note that the market for such tools is quite extensive; the election is also varied and it is your choice. In any case, it’s better not to save on statistics as well.