For a more comprehensive list of monitoring tools and their features, check out this Wikipedia page.
As the question states, what are the most commonly used tools used for this task and what are their strengths and weaknesses?
Server Fault is a question and answer site for system and network administrators. It only takes a minute to sign up.Sign up to join this community
I'm using PA Server Monitor . It's primarily Windows focused (event logs, performance counters, services, etc) although getting better with other systems now that some limited SNMP support has been added. The thing I like best is it's easy to configure compared to a lot of apps (no config files, no command lines, etc). I wouldn't recommend it for a heavy *nix environment though.
Oh, it's not free, but less expensive than some competitors.
I use Polymon and love it.
It's fantastic for monitoring anything that can be communicated by TCP Port, SNMP, Powershell, WMI, SQL, HTTP, Perfmon, or Ping.
I don't monitor anything *nix, so I can't speak to that. But for the Windows world it's very simple to set up, extremely intuitive, and extremely flexible, It has very nice built-in dashboard display, sms or email notification, etc.
We needed something customisable as we need to monitor some systems which are not online all the time, but can send mail or be dialled in.
We tried nagios (maze of scripts), AppManager (nice, but nonadaptable), Zenoss (nice, but when you mention Oracle, price gets hefty multipliers) and landed on Zabbix which has open protocol, open database structure, heck, I can write a plugin on every level in a hour. It's nicely compartmentalised (server, client, database, ...). And it's web frontend is quite nice and customisable.
YMMV, for us the monitoring of "offline" systems is important and it is usually not covered by such software.
I've worked with Pandora FMS, and I like it mainly because it's very flexible and easy con configure for the average sysadmin. Also I like the web interface with all the reports and the extensive documentation. And not very useful for a single datacenter, but very cool is the geolocation interface that shows the position of the agents monitored.
I've also tryed Nagios and I like all the plugins it has, and that it's well known among sysadmins.
Note: I've been one of the developers of Pandora FMS for some time.
We use Level Platforms for this task. Provides a ton of useful information without overloading the sysadmins, and makes it extremely easy to handle all of the hardware in our server room (as well as many of our clients').
Also take a look at Argent Guardian. It's cross-platform, can function as a syslog server, they'll give you the database schema to do your own reporting, if you need that, and you can import your own images as "maps" to give visual alerts.
For the status of servers and services (whether they are up or down, and sending warnings if they go down) and for yes/no questions ("has a backup been done in the last 24 hours?") we use nagios. It is hard to set up, but it is immensely configurable. Custom scripts can be run on remote computers. Alerts can send emails, send text messages or even run custom scripts.
For the health of servers we use munin - it provides nice graphs of memory usage, cpu usage, network usage etc. Pretty easy to set up on linux at least (I have not tried with Windows).
I've used Activexperts Network Monitor with great success (on a mostly Windows network but it had some unix and linux hosts, printers of various brands and so forth that was also monitored with it).
It's really easy to setup and learn, rather cheap for what you get (was $500 for site/enterprise license) and supports vbscript and remote unix commands. If the network is small (a few hundred nodes at most) I think this is much more intuitive than System Center Operations Manager which feels more directed at huge windows networks only.
Network Monitor comes with a lot of predefined scripts for monitoring stuff like e-mail servers including various Exchange versions and all its services, http servers with expected response, event logs, sql queries and expected responses and so on.. .and dependencies are easy to configure ("all these depend on this router so if it fails to respond to ping and snmp, don't bother alarming us about all the stuff behind it that's not responding"). SMS with gateway or local GSM modem support and all rules can of course have actions like service restart, server restart or custom script - to fix reoccuring problems for you (it's important I think, kinda like regression testing is for development).
...I've also tried to tame a Hobbit and didn't really enjoy it at all (nor the bloated Windows agent) - but it was set up for Windows server monitoring and it really blows at that - most likely more suited for a linux or unix-centric network.
We use hyperic - it has both an open source version and a commercial one
It monitors the operating system (RHES 3, 4 and 5 + Ubuntu), Apache, MySql, JBoss, Tomcat, mail servers, memcached and it probably can monitor more applications. No special configuration is needed, all servers were found with the auto discovery, even if they were installed in an untraditional place. It is very easy to use and configure, you can control your services (start/stop etc.) and define alerts.
Minuses - You need to configure it to run on boot (we are using cron to do that).
I notice no one has mentioned HP SiteScope yet
We use WhatsUp from ipswitch, it's very easy for setup small networks, it can autodiscover networks by port scan, it can use windows and SNMP credentials.
For monitor statics like cpu, mem, and disk, we need to setup SNMP. WhatsUp support SNMP v1, v2, v3.
WhatsUp have a passive monitor through syslog (Unix), event viewer (Windows) and SNMP Traps.
It has a nice ajax web interface with custom user and custom workspaces.
P.D. sorry for my bad english
I've used hobbit, big brother and nagios when working for poorer (read cheaper) organizations. Of the three I prefer hobbit because its simple and bulletproof. I've always felt that nagios is is trying to be an open source version of openview or tivoli, and frankly if I have the time to spend configuring a framework like openview or tivoli then monitoring is probably my entire job and my organization can probably afford to buy openview, so why use nagios?
We've just started using "Servers Alive" which is very inexpensive, it isn't too pretty looking, but it supports a tonne of different checks and can alert in several ways, handles technician scheduling/rosters etc for any notifications. You can also make checks rely on others, i.e. "this" system requires "that" to be up/running.
We use Orca to monitor our systems. It's not super pretty, but it gives a ton of low level details other monitoring systems don't use.
I use a combination of Nagios, Cacti, custom scripts and one of my own projcts -- System Health Monitor. I like having external service monitoring as well as graphs of system resources so you can do post-mortem analysis of system problems or quickly check the graphs to see if things look 'normal' compared to their historical values.
I am using nagios and hobbit (bigbrother opensource implemenation) independantly and have found both having positive and negative qualities.
pro: has a nice sub-minute scheduler for running tasks at regular intevals and has an embedded perl interpreter to boot.
con: config insists on having a 'server' for every test, when sometimes you just want to run a test that is based on an application 'feature' but not necessarily isolated to a single host. Revert to a meta-config that generates the actual nagios config to overcome this.
pro: opensource compiled server instead of the massive scripts used by original big-brother easy integration with the bb client 'dboard' command to poll data.
con: also stuck in a 'server-oriented' mentality, which fits most folks, but not me.
We us IP Check which has been renamed PRTG it allows for a wide range of sensors that can monitor all sorts of different activity.
I use NetGain Enterprise Manager from NetGain Systems. It's take just few minutes to install and get it up and monitoring. Best of all, it's free. check out http://www.netgain-systems.com
I've worked with a lot of monitoring systems at a lot of places. Most of them have already been mentioned. Here are a few that haven't been:
SMARTS - now owned by EMC. Really is the best thing ever for root cause. It's not cheap and support may not be good anymore as it's owned by EMC. We were lucky enough to work with the founders of the company to get it implemented.
Big Brother. Nice and simple, but a bad license. It's also the ugliest web gui I've ever seen, so I had to rewrite it. Never got Big Sister to work.
HP Openview, when engineered, installed and run by a competent engineer can be good. However I've only seen it done right once and wrong more often than I can remember. I would never choose to use it.
BMC Patrol. Just awful. Die, die!
And finally, for logs and tracking down problems you just have to use Splunk. If this had been around 10 years ago I would have saved myself a lot of wasted time.
We've tried Applications Manager Its running on java and mysql. It's really powerful and easy to configure from the browser. It's not that expensive either.
Currently we use SCOM from MS. I wouldn't recommend it to anyone!
We use Ipswitch whatsup gold 12 for monitoring about 2000 devices, both performance and tcp/ip or wmi based monitors and both windows and linux. Good thing about it is that it is easy to use and configure, has bulk change options and autodiscovery, multiple notification methods. The bad side: seems to have had a limit of about 2000 devices, after that performance was getting slow, plus it only runs on windows. The distributed version doesn't really deserve the name and the price tag. We evaluated nagios (setup too complex for a dynamic environment), zenoss (no bulk change or autodiscovery, too limited for dynamic environment) and currently looking at Zabbix, which seems most promising with all the nice features Whatsup has and more, such as fully distributed architecture with probes and server, relatively simple setup, open source backend (mysql, apache)...
I've been using Sysmon for a number of years. There are a few modern services that it doesn't monitor, but it compiles easily on most *nix platforms, has almost no dependancies, is extremely light-weight, can monitor very large numbers of devices and services with ease, can handle complex network layouts (incl. ring topologies) and failover monitoring. It's basically a config file deal, but the format is pretty easy (based on plist/css).
Nagios and HPOpenview are the two that I am familiar with and have experience in. Both are good choices, although for the latter I'll echo other posters that it needs someone that knows how to do it right. hen again the only place I saw it running was when I was with HP so that might have helped my perception.