At the risk of getting off-topic, I'm tired of stuff that doesn't quite work. (can't comment on the actual book because I haven't read it)
However, I can't see how Nagios can even begin to satisfy the needs of most modern IT operations folks. These days, most people need to know a lot more than whether machine X is up. They need to know which part(s) of their web apps are not functioning correctly. They need a lot more intricate detail than is possible with Nagios or SNMP-based monitoring tools. Really, the only monitoring tool that does it for me is Hyperic.
Look, everyone likes to talk about agentless, and they're full of crap. Everything runs an agent - your agent might be SNMP, or whatever Windows happens to run, but you're running an agent whether you admit it or not. Now the question becomes, why run a crappier agent? Why not run an agent that actually gives you data on your applications? Everybody knows that your service/server is "working", but how *well* is it working?
Slashdot Mirror
Very good points. Looks like this article agrees with you - http://searchenterpriselinux.techtarget.com/origin alContent/0,289142,sid39_gci1250897,00.html
At the risk of getting off-topic, I'm tired of stuff that doesn't quite work. (can't comment on the actual book because I haven't read it) However, I can't see how Nagios can even begin to satisfy the needs of most modern IT operations folks. These days, most people need to know a lot more than whether machine X is up. They need to know which part(s) of their web apps are not functioning correctly. They need a lot more intricate detail than is possible with Nagios or SNMP-based monitoring tools. Really, the only monitoring tool that does it for me is Hyperic.
Much easier to set up and get running - http://www.hyperic.com/ Not to mention supports more platforms than all of the others.
Either that or an employee submitted it ;) Yo Mark, is that you?
You may want to try Hyperic's VMware management - http://www.hyperic.com/products/managed/vmware-man agement.htm
Look, everyone likes to talk about agentless, and they're full of crap. Everything runs an agent - your agent might be SNMP, or whatever Windows happens to run, but you're running an agent whether you admit it or not. Now the question becomes, why run a crappier agent? Why not run an agent that actually gives you data on your applications? Everybody knows that your service/server is "working", but how *well* is it working?
Besides, for real systems management, see Hyperic.