You can monitor load on virtual machines in the cluster over a period of time in such parameters as the following:
- inbound traffic,
- outbound traffic,
- input-output operations,
- CPU time (number of instructions).
Load statistics is collected every 5 minutes. You can choose a period for statistics display (previous day, month, any desired range, etc.), an interval (days, hours, minutes), a cluster node, and virtual machines (you can choose any specific server, all of them, or top-10).
Cluster administrators can monitor resources and loads on all cluster nodes. Resources monitoring collects statistics on RAM, HDD, and CPU usage, while load monitoring gets data on inbound and outbound traffic, CPU load, and input-output operations.
Parameter statistics is collected online and can be viewed with 1 minute detalization. In real life it allows a VPS provider to determine which user overloads RAM or chokes up the traffic channel, thus causing problems for its cluster node neighbours. By using this information the administrator can easily make a decision about migrating the virtual machine in question to the other physical server.