Google Cloud customers will now be able to suspend their virtual machines (VMs) when not in use which will help lower their cloud spending.
The software giant’s cloud computing division has announced that its new Suspend/Resume feature for VMs is now generally available after launching in alpha several years ago.
The new feature works in a similar way to closing the lid of your laptop or putting your desktop PC to sleep. By suspending a Google Compute Engine VM, the the state of your instance will be saved to disk so that you can pick up later right where you left off.
The best part about Suspend/Resume in Google Cloud though is that customers will no longer need to pay for cores or RAM when their VMs are in a suspended state. However, they will still need to pay the cloud storage costs of their instance memory as well as other VM running costs like OS licensing but these may be reduced.
Suspending and resuming VMs
When a Google Cloud customer suspends an instance, an ACPI S3 signal is sent to the instance’s operating system just like when you close a laptop’s lid or put a PC to sleep.
The company makes the case that using this type of signal allows for broad compatibility with a wide selection of OS images so that customers don’t have to use a cloud specific OS image or install daemons. At the same time, undocumented and custom OS images that respond to the ACPI S3 signal may also work with Google Cloud’s Suspend/Resume feature.
It’s also worth noting that storage is dynamically provisioned when Suspend is requested and is separate from the instance’s boot disk. Other cloud providers require users to ensure that they have sufficient space in their boot disk to save instance states which may increase the costs of running VMs.
In a new blog post announcing the general availability of Suspend/Resume, Google Cloud also pointed out that the feature can be used by organizations to deal with demand spikes as they can initialize instances with their critical applications and then suspend them so that they can be resumed later. Although Compute Engine instances can be created quickly, resuming an instance is much faster than creating an entirely new instance.