Containers are rapidly becoming a must-have element in the modern enterprise IT toolkit. As Dave Raffo noted in his Technical Insight Report on Container-Native Storage (April 2021), the drive to digital transformation, along with an increasing shift to hybrid and multi-cloud deployment, is creating rapid growth in the use of containers. 2020 research by Evaluator Group indicated that over 60% of IT infrastructure clients were considering or using containers in production.
Rapid proliferation of containers creates a need for automated orchestration, and so the market is seeing any accompanying growth in the adoption of Kubernetes, which has become the de-facto standard orchestration platform.
But containers are not needed to address every problem or every use case. Let’s examine where this technology solution fits in a successful enterprise technology strategy.
A discussion with any DevOps team will overwhelm an IT executive with a flood of discussion regarding virtual machines, containers, and Kubernetes, but may still leave the listener wondering: what are containers good for? Why would I choose to deploy a container instead of (or in addition to) a virtual machine?
The apartment building versus hotel room metaphor is an excellent model for comparing these two technologies.
An apartment building shares a common infrastructure (hardware) amongst a large number of individual units (apartments), each of which can be decorated uniquely (by the tenant) within its designated four walls. A tenant engages with the landlord for use of the apartment. The tenant must contract for and decorate his space, move in, stay for a period of time but then perform work to return the space to “normal”, release the space and move out. This overhead of the rental process can be burdensome and time-consuming.
A virtual machine works in much the same fashion. Virtual machine solutions enable the sharing of a physical server (or set of servers) amongst multiple virtual machines (like the apartments in the building). Because each virtual machine has its own operating system, it can be unique. The “landlord” in this model is a “hypervisor” which translates between the physical hardware and the virtual machine’s view of the universe (see Figure 1, below). But like an apartment building, the time for individual tenants (VMs) to provision and move in (or give notice and move out) can be extensive…
Download now to read the free Technical Insight report.