HCI vs Private Cloud vs Public Cloud for Servers

HCI vs private cloud is not a like-for-like choice, because the two terms sit on different layers. However, the real choice is where each workload runs, and five rules about your own workloads and people settle it, as set out below.

Before the comparison, here are Signisys’ working descriptions. Hyperconverged infrastructure packages servers, storage and networking into one cluster, while a private cloud delivers self-service services dedicated to one firm, and a public cloud delivers services to many customers from a provider’s estate. This article is for the infrastructure lead of a mid-size firm based in India or the GCC that must pick a home for the next servers, and for the CFO or business owner who approves the pick. If aging hardware is what prompted the question, why aging hardware forces the question sets out the risks. This article also has no build steps, no scoring and no prices.

HCI vs Private Cloud Compared: Building Block or Operating Model

Before comparing, note the one difference that frames the whole comparison. HCI is a way to package hardware, while a private cloud is a way to run and request services. A private cloud can sit on HCI or on other hardware, so the HCI vs private cloud question is about layers, not rivals.

Decision Rules for HCI vs Private Cloud vs Public Cloud

Before you pick, five rules follow, each a question about your own workloads and people. The wider private cloud vs public cloud debate becomes easier when each workload has its own answers.

Across the five topics of workload, growth pattern, skills, location and control, a yes leans toward keeping the workload on your own site or in colocation, while a no leans toward testing it in a public cloud or with a partner-run private cloud.

HCI vs Private Cloud Rules 1 to 3: Workload, Growth Pattern and Skills

Rule 1, workload: does it need to sit close to a plant, a branch, a machine or data held on site? A yes points to on site or colocation, while a no leaves room to test elsewhere.

Rule 2, growth pattern: is demand steady enough to buy capacity ahead, or does it swing? Steady demand leans toward capacity you place yourself, while demand that swings leans toward testing a public cloud.

Rule 3, skills: does your team already run a virtualization and automation layer day to day, with named people? If the answer is yes, a platform you run yourself is a fair choice. If you can name no one, a partner-run private cloud or a public cloud is worth testing.

Also, the dated records of support ends and recovery tests give the evidence behind these answers.

Rules 4 and 5: Location and Control

Rule 4, location: which rules, customers or response times tie this workload to a country or a site? Ask the compliance lead. Also, India and the GCC states each have their own requirements, and this article makes no legal claim about them. For firms weighing HCI vs private cloud vs public cloud, the compliance lead’s written answer comes before any platform name.

Rule 5, control: does your firm need to choose hardware, patch windows and access paths itself? Ask the IT head. A yes leans toward a platform your team runs, while a no leans toward a service where the provider or partner makes those choices.

When to Choose HCI vs Private Cloud on Your Own Site

When the rules point to your own site, the question changes from where to which layer. Then three questions, as a Signisys suggestion, settle it.

1. Are you buying hardware, or changing how teams request services?
2. Do you keep hardware you already own?
3. Who requests servers today, by ticket or by self-service?

Use this as your own decision test, but not as a claim about products. If the answer is a hardware decision, HCI is the layer to look at. If it is a service-model decision, a private cloud is the layer to look at. Both can apply, and HCI can then be the base. A private cloud can also run on hardware you already own.

If your servers sit across more than one of these layers, that mix has a name: a hybrid estate. The next section then sets out how to run one without a single platform for everything.

A Hybrid Path: Mixed Estates and One Layer Per Workload

However, a firm does not need one platform for everything. Instead, each workload names one layer and a review point. Also, a government definition of private and public cloud keeps the terms straight. In NIST SP 800-145 (September 2011), a private cloud is provisioned for exclusive use by a single organization and may exist on or off premises, and a public cloud is provisioned for open use by the general public and exists on the premises of the cloud provider. This article names a layer for a platform decision, but it sets no target and no mark.

Three Questions Before a Workload Changes Platform

These are Signisys’ suggestions, but not a standard.

1. Do the five rules still give the same answer for this workload?
2. Before the change, does the person who holds the workload’s budget agree in writing?
3. Does a route exist, if needed, to bring the data back out?

If all three answers are yes, change that workload. If any answer is no, the workload stays where it is, and a named person owns the gap. Before any build starts, the build order after you pick a platform sits in a separate guide.

Review Points for a Mixed Estate

Give each placement a named review date. Two things bring that date forward: a change in the rules that tie a workload to a site, or a change in its demand. These are decisions for finance and the IT head, while inventory and patch review dates belong to the refresh plan and fall outside this article.

Who Signs Off HCI vs Private Cloud vs Public Cloud

Four roles each own one decision about placement. A workload moves only when each role has answered.

Before a workload moves, the CIO or IT head names the layer and who runs it, and the evidence to request is an owner by name and a layer in writing for each workload. When the firm plans spend for that layer, the CFO or finance controller agrees how, and the evidence is the finance lead’s written agreement on the planning method. Also, the security or compliance lead answers the location and access questions, and the evidence is answers in writing to rules 4 and 5. Finally, the workload owner in the business accepts the platform change, with a written acceptance naming the workload and the date.

This is an offer: if you want one, Signisys provides consulting and managed services in this area, and we can give you a second opinion on your platform choice. It also draws on our design and delivery for on-site and cloud estates and our network, hosting and data center support. No charge, with a fixed scope.

HCI vs Private Cloud FAQ