how do you measure your server load (let's concentrate on nginx and Apache maybe)?
What metrics are you using (cpu usage, site load avg?)?
When and how are you deciding that you need more powerful server or a cluster of servers?
Any help and suggestions will be appreciated.
The main metrics I use are CPU usage, swap usage, HDD usage, and network card usage.
Could you tell us how do you measure that metrics?
If your server is Linux based then some common options are:
Then you need to ensure that your Raid controller cards are keeping up, so you need to ensure that your RAID card or soft RAID have proper configuration tools installed and preferabbly emails you when they are taxed too heavily.
Coincidentally, if you have access to a virtual server then any of the aforementioned servers will allow you to manage Windows, Mac or Linux network monitoring using SNMP. They support open source RRDTool's graph items, so as long as you have either a virtual Linux or a Linux box designed for exclusively this functionality, you'll have a much better handle on when it is time to replace or add more hardware, manage space and set throughput thresholds (that notify you when reached).
There are additional things that can be done, however this is a lot to get start.
There isn't any one "solution" to a question like this. It all depends on how you have your server set up.
Thank you ServerStorm! That was very helpful.
This answer pretty much sums it up. I would also add Hyperic HQ and Groundwork as potential choices. This article does a nice job breaking down some of the options. http://www.futurehosting.com/open-source-server-monitoring-what-are-your-options/
This topic is now closed. New replies are no longer allowed.