On 23-Nov-2012, at 7:37 AM, Train Lau <szliuq...@gmail.com> wrote: > Jerry > > Can u introduce some open source products about the agent-type monitor. >
There are potentially 3 things major systems one could monitor: 1) The hypervisor hosts 2) The VM containers 3) The resources consumed by the user's OS running inside the VM container (business layer) As a cloud infra admin, I would be responsible for (1) and (2) but not (3). As for (3), if you treat what's running inside the VM as just another Linux OS, there are many monitoring platforms that one can use to track host metrics. Nagios, Ganglia, Cacti, Collectd are some of the popular ones that's shipped with Ubuntu LTS. Ganglia+Nagios could serve the requirements for (1)+(2) and (3). You could spin up one monitoring deployment which takes care of the cloud admin's monitoring needs and another deployment specific to the business layer needs of the running OS inside the VM. Keeping them separate will eventually allow to transition the user's monitoring requirements as a self service layer which can be easily puppet automated or templatized during VM creation itself. Most of the time, the monitoring deployment plans really depend on the team structure and ownership. Hth @shankerbalan
