Google Introduces Single Tenancy VMs On Compute Engine

Google Compute Engine, the IaaS component of Google Cloud Platform, introduced dedicated physical servers that can run VMs from a single tenant. This new placement strategy will help customers in meeting compliance and regulatory requirements.

Infrastructure as a Service is one of the key cloud computing delivery models. Hyperscale cloud providers such as Amazon, Google, IBM, and Microsoft invest in physical hardware that exposes the core components of infrastructure – compute, storage and networking – to customers. Compute is often delivered through virtual machines that are placed in one of the physical servers.

When a customer deploys a VM, he doesn’t control the placement logic which means that it will be provisioned in one of the physical servers based on available capacity and utilization pattern. The scheme in which different VMs belonging to different customers run on the same physical machine is called the multi-tenancy model.

While multi-tenancy delivers better efficiency and economy of scale, a certain type of workloads suffers from the noisy neighbor problem. A VM running in multi-tenancy mode can consume maximum resources depriving other VMs their share. This scenario results in degraded performance of applications that demand unmatched bandwidth, storage throughput, and CPU time.

Read the entire article at Forbes

Janakiram MSV is an analyst, advisor, and architect. Follow him on Twitter,  Facebook and LinkedIn.