2

I need a monitoring system, much like ganglia / nagios that is build for the cloud.

I need it to support :

  • Adding / removing nodes dynamically. (Node shuts down, dose not imply node failure...)
  • Dynamic node based categorization, meaning node can identify them self as being part of group X (ganglia gets this almost right, but lacks the dynamic part...)
  • Does not require multicast support (generally not allowed in cloud based setups)
  • Plugins for recent cool stuff such as Hadoop, Cassandra, Mongo would be cool.

More features include: External API, web interface and co.

I've looked at Ganglia, munin and they both seem be almost there (but not exactly). I would also go for reasonably priced Software as Service solution.

I'm currently doing research, so Suggestions are highly appreciated.

Thank you,
Maxim

Maxim Veksler
  • 2,555
  • 10
  • 27
  • 32

2 Answers2

2

I was going to suggest ganglia but I see you already considered it. There's also Icinga and Reconnoiter, that one being a very new addition. Reconnoiter has a hosted version with an amazing interface, called Circonus.

coredump
  • 12,573
  • 2
  • 34
  • 53
0

Really what you are talking about is some sort of config mgmt system to manage your monitoring. In essence you need monitoring to be provisioned when you provision a new host and deprovisioned when you shutdown a host. In house we are using puppet to provision our hosts into nagios and then manually removing as that is not a common task. As time goes by the deprovision process will be automated as well.

I would look at chef and puppet as the top contenders in this space.

cwebber
  • 491
  • 3
  • 7