On our VPS platforms, we use automatic migrations to move VPSs from one hypervisor/server to another. This has two major advantages:
- We can update our platform without you noticing
- With a high CPU load of a hypervisor, we can automatically migrate VPSs to a server with a lower load
In addition, the same technology also allows us to perform manual migrations, for example to move a VPS to another hypervisor if the original one becomes unavailable due to an incident.
During a live migration, the memory of your VPS is first replicated to the destination server. Once the differences in the memory on the source and destination servers are small enough, your VPS will be paused and transferred to the new server within a second.
-
Steal time due to shared CPU load does not apply to PerformanceVPSs: the CPU cores are reserved for you.
- Is your VPS (and not the hypervisor) experiencing a high CPU load? In this tutorial we explain how you can investigate the cause.
Automatic migrations based on CPU load
Most VPSs only use part of their CPU capacity. For example, web hosting requires proportionately more RAM than CPU power. However, the longer a VPS is in use, the more intensively it is usually used, for example because you process more web traffic after X time. In addition, the CPU load on a hypervisor can increase because more VPSs are delivered on the same hypervisor. The consequence of this increasing CPU usage is that the CPU load of a hypervisor increases over time. In the worst case, steal time can arise.
We therefore automatically monitor both the CPU steal time and CPU usage of all our hypervisors. When the CPU load or CPU steal time reaches a certain percentage, automatic migrations are started to move VPSs to a hypervisor with a lower load. This way we prevent the load or steal time from increasing, or that, for example, 100% of the CPU cores of a hypervisor are used.
This brings us to the end of this article about automatic migrations on our VPS platform.