kentoh - Fotolia

How do I manage VM resources in hyper-converged systems?

Hyper-converged systems are growing in popularity, but the setup of virtual environments can result in wasted VM resources.

In hyper-converged systems, virtual machine (VM) resources such as compute and capacity cannot be scaled independently of one another. This can sometimes be problematic because business workloads vary. Hardware that is appropriate for one organization might not be adequate for another.

Scalability can sometimes present a problem because of resource contention. In a virtual server environment, all of the VMs that are running on a server are competing for physical resources such as CPU cycles, memory and storage I/O. This sharing of resources works fine so long as each VM can get the resources it needs to run its workload efficiently. Problems only begin to occur when VM resources become scarce and virtual servers must start competing for access to hardware resources.

In a typical virtualized environment, organizations usually address resource contention by adding more hardware resources. In a hyper-converged infrastructure however, that may not be an option depending on which vendor's solution you are using. That being the case, there are some things you should do prior to investing in a hyper-converged system:

  • Assess your hardware requirements as accurately as possible. Don't just evaluate what you are going to need today, but try to figure out what VM resources will be required to sustain the needs of your virtualization infrastructure over the next few years.
  • Choose a product in which the storage is equipped with its own dedicated CPUs. That way, these dedicated CPUs can handle the storage-related tasks rather than robbing the servers of CPU cycles. Some of the storage-related functions that may benefit from dedicated CPUs include replication, acceleration, deduplication and even copy jobs.

If a hyper-converged system is not able to offload some of this heavy lifting to the storage hardware, then server CPU cycles and server memory will be wasted performing storage-level functions. This places a load on the server hardware, decreasing available VM resources.

Next Steps

How to strike the correct VM resource allocation mix to meet performance needs

Is hyper-converged storage a viable SAN alternative?

Hyper-converged systems gain traction in the market

Scale-out capacity with hyper-converged storage systems

Dig Deeper on Hyper-Converged Infrastructure Management