The advantages of one vs another monitoring service is if it does server level monitoring (CPU, Mem, disk usage), application level (sends information to your web app and expects a certain response), or both.
As previously mentioned, you should have both. If your app is working well, chances are that a system level alert will be triggered by high traffic spikes, etc and you can fix the problem before your applications suffer. But application level alerts are the most important, because they typically test what a user sees and can alert you when any part of your web stack has an issue (apache goes down, dead links, broken embeds, database corruptions, etc, etc).
Sophisticated services like New Relic (NR) can test both.
Regarding your second question (is it worth it), my opinion is that it depends. I have been using New Relic for a few months and find it quite useful. Especially on determining why a problem is occurring instead of simply receiving an alert. The alerts are sophisticated and NR can monitor both applications and the server itself. Overall, I find NR to be a great product for total server and application monitoring, but it is expensive. But for a small server (such as mine that receives only 2000 visits a day across 10 small websites and generates little revenue), I am not finding justification for the cost. I m planning on switching back to free/more affordable services for basic alerts and leaving the troubleshooting up to me.
For free basic services, I use cloudkick s free developer plan for server monitoring and the less complex/costly pingdom.com for 1 free application monitor. Pingdom s alerts are not as sophisticated, but it does basic application level monitoring and alerting at reasonable prices.
My suggestion would be, if your uptime is critical then you should include in your budget a sophisticated monitoring system like NR. If your uptime isn t critical then a few free services may do the trick. But either way, you should be monitoring both your server and applications.
