Original URL: http://www.theregister.co.uk/2013/04/25/rackspace_server_monitoring/

Rackspace plugs event monitoring into its servers

Expands cloud monitoring tech with 'Virgo' agent project

By Jack Clark

Posted in Cloud, 25th April 2013 00:46 GMT

Rackspace has broadened the capabilities of its cloud monitoring technology by letting admins put an agent on their servers to get a clearer idea of how apps are using hardware resources.

The upgrades to Cloud Monitoring include server monitoring, a graphing feature for viewing old data, PagerDuty integration, and multi-data center redundancy, Rackspace announced on Wednesday.

The Server Monitoring technology works via a monitoring agent named Virgo, which lets admins perform checks on load average, CPU, memory, filesystem, and network usage. Admins can design their own checks via the API.

If you put all this together, the tech would theoretically let an admin write a script for querying the number of active sessions on a database, which the agent could then execute at pre-determined intervals, alerting the admin if the number went over a pre-agreed "safe" amount via paging them through PagerDuty.

The agent works across both Windows and Linux cloud environments, and is built on top of Node.js's libuv cross-platform library. The overhead of the agent is around 6MB of RAM per cloud server.

Rackspace have published the agent's code as open source via the "Virgo" project on Github. "Virgo is a project for building an on-host agents. The goal is to provide shared infrastructure for various types of agents," the company wrote on its Github page. "Join in and build your agent with us."

Server Monitoring will be free until July 31, the company said, though it has yet to disclose pricing details thereafter. Much of its technology is based on tech from monitoring outfit Cloudkick, which Rackspace acquired in December, 2010.

Though the tool will be useful for Rackspace admins, it also highlights the loss of information that happens when people move to the cloud. If you operated your own servers within your own data center, then you'd have access to all this information anyway, though it would require a bit of fiddling.

Google engineers may scoff at the tech as well, as the Chocolate Factory is currently rolling out an experimental monitoring technology named CPI2 that inspects every single task on every single one of the company's servers and can analyze performance down to the individual CPU level – but whether a typical cloud admin would want to deal with task-specific latency reports on individual chips is another question. ®