The CloudHealth Agent is deployed to running cloud instances to obtain instance level metrics for CPU, disk and memory, which can be leveraged for rightsizing.

The agent works by taking snapshots at specified intervals and reporting them to CloudHealth on an hourly basis. Many CloudHealth customers also use CloudWatch, which provides CPU level metrics. The CloudHealth Agent can be used to provide a complete picture of your instances by also delivering disk and memory metrics. It is lightweight and requires only 20 MB when running and will not increase. When idle, the data is already transferred to CloudHealth and does not require local storage.


The lightweight agents has a minimal impact on instance resource usage. It aggregates data every hour and sends it back to CloudHealth and then goes back to sleep. The chtcollectd daemon, which is our custom wrapper around collectd, is set to collect metrics every 10 seconds. These metrics are then stored to disk. Nothing is held in memory. When the agent is running, it only uses single digit to a very low two digit MB of RAM.


It transmits data back to the application by pushing data from the instance (over https - SSL encrypted) once every hour to the CloudHealth endpoint. A JSON file of the aggregated Disk I/O, Filesystem, and Memory metrics are transmitted (pushed) by the CloudHealth agent.


The agent enacts Disk I/O, RAM & CPU by doing the following: metrics for Disk I/O, RAM & CPU are by default collected every 10 seconds (can be configured) for the instance and transmitted once every hour to the CloudHealth endpoint. The default settings are visible on this page:


The agent leverages collectd but is fully packaged in a way where it should not and does not conflict with any other software running collectd. We have specifically designed and packaged the agent to avoid conflicts. In the rare circumstance the agent has a conflict, we will be happy to investigate and resolve accordingly as a priority.


CloudHealth offers both Linux and Windows agents. Supported OS servers include: Ubuntu 12.04 and above, RHEL 6.5 and above, Amazon Linux 2014.09 via RPM or DEB packages, and Windows Server 2008 R2 and above.


There are no maintenance requirements for the agent unless you turn off Auto-Update. By default, the Agent checks in with the CloudHealth endpoint to see if a new version is available and auto-updates, though version updates are uncommon.


The agent collects data from instances and saves it to the disk, so there is no memory: overhead. In case the instance is unable to interact with the CloudHealth endpoint, metrics older than 24 hours are discarded.