I)Introduction
This is something that I’ve been thinking about since Broadcom bought VMware and injected a lot of uncertainty in to the world. I’ve personally seen what a renewal in the Broadcom Era looks like and it isn’t something that is compatible with the business practices or budgets of many organizations.
My focus at this point is in the native, on premises hypervisor and base level management. To put it in product terms, vSphere and ESXi. There are a huge amount of businesses that use VMware without bolt ons, which is one of the reasons why the new SKU options offered by Broadcom are causing companies to re-evaluate their choice.
Navigating a hypervisor swap is an intricate and nuanced conversation. This is some base line data based on my research. There is nuance to all of this data and how it applies to users. This information will likely also be useful to those who want to scale up or down an IT footprint.
II)Business situations and requirements
There are three key scenarios that I am going to delve in to. Each has its own challenges. These are generalities and the business needs to weigh costs and risks accordingly.
- A small remote office/Back office site. This is generally a server where cost is a big driver and risk of compliance or cost are low if the system goes down or data is lost. An example may be a file server at a small site. The site may have a few physical servers and/or a small number of VMs but they will have little redundancy on premises. Replication to data center, a large remote office/back office site (hub-spoke) or a cloud is an option here.
- A large remote office/Back office site. These sites are more consequential in nature. Redundancy comes into consideration. They may have a few dozen VMs including a database backed application stack. Infrastructure would look like a storage array with a few servers connected to it or hyper-converged configurations. Switching may be shared between storage and other traffic. They’re designed to be fault tolerant in a way that a small site wouldn’t, but don’t have scale and some dedicated infrastructure a data center would.
- Data center environments. Some data center environments may look very similar to the large Remote Office/Back Office configurations. The differentiation at a hypervisor level to me is Fibre Channel storage and/or a dedicated storage network. This offers better isolation against noisy neighbors and simpler security practices overall.
These are not hard absolutes. A large Remote Office/Back Office environment may look like a data center environment, and a small data center environment may look like a large remote office/back office environment. Existing infrastructure may well dictate architecture.
III)Bake off criteria
There are a lot of parameters to evaluate in a hypervisor. Although business dynamics drive the current market flux, cost is a secondary driver to the technology being adequate to perform clear goals. In my mind these should be as follows.
- The organization has skills or can train readily to understand the platform, or contract with people who do.
- A clear transition path to the new system from old systems.
- Capability to run applications that the organization needs to.
- Support of connectivity to facilitate use of the applications the organization needs to.
- The ability to maximize current and previous investments in infrastructure.
- Support quality. This includes things like documentation.
- A licensing and support model that the business can live with.
- Cost.
IV)Finally, the data points
This is deep into the weeds. As a technologist it will likely be where you want to look. If you’re in the management space, it’s something that you may want to turn over to the technologists running the environment. Business cases are extremely useful to call out things evaluated in the Hypervisor Bakeoff Criteria Spreadsheet. Each hypervisor will have a comments section and an evaluation section. The comments are notes I’ve taken for myself about working within the systems and may be useful in explaining some grades.
V)Status
This spreadsheet is constantly evolving. I’m currently working with Proxmox VE, so see my other posts related to my configuration. In the candidates section are the hypervisors I’m wanting to work with. The grading process is more feeling based than rationalized currently, I am working on trying to determine criteria to use in order to cleanly grade.
VI)Future revisions and derivatives
As of now the grading is not based off specific criteria. I would like to define that criteria. The end goal of this is to do a comparison of how these hypervisors compare to VMWare, which is considered the hypervisor that defines the industry currently.