This content is part of the Essential Guide: Private cloud implementation guide

Four private cloud design pointers to ensure success

To design a successful private cloud, lay the groundwork: Plan beyond the five-year mark, determine who will manage images and make it highly available.

Building a private cloud is anything but trivial, but there are some private cloud design pointers that can help make implementations easier.

Like public clouds, private clouds can take many different forms. On one end of the spectrum, there are enterprise-grade, infrastructure-as-a-service clouds that rival those of public cloud providers. On the opposite end of the spectrum are vendors that market consumer-grade, external storage devices as private cloud storage. Ultimately, there is no completely rigid definition of what constitutes a private cloud, although many organizations define private clouds as self-service, virtual machine (VM) provisioning environments.

Define objectives

One of the most important keys to successful private cloud design is to define your objectives upfront: decide which features and capabilities you will need. It's also important to estimate the workload that you expect the private cloud to handle, and to define any software compatibility requirements.

Creating this blueprint will help you develop a private cloud architecture that is well-suited to your organization's needs, and it will help keep your budget in check. You're less likely to fall victim to vendor upselling if you already have a well-thought-out plan in place.

Design for the future

Many organizations build private clouds that are estimated to meet their needs over the five coming years. This is a reasonable approach, but there are two additional aspects that you must consider when designing your private cloud for the future.

First, it is important to have a strategy for increasing the cloud's capacity if necessary. Suppose you build your private cloud on hyper-converged infrastructure. Adding capacity is just a matter of installing more nodes, but organizations must consider whether compatible nodes will exist in a few years. If not, it's important to include a solution for scaling the cloud's capacity using alternative hardware in your private cloud design.

Second, you must create a plan for transitioning your private cloud into a hybrid cloud. This is an important consideration, even if your organization has no intention of creating a hybrid cloud. Experience has shown that nearly all private clouds eventually evolve into hybrid ones. The transition will be far smoother if you account for it during the initial private cloud design phase.

Delegate template maintenance

Many private clouds exist as self-service environments, where authorized users can deploy VMs on an as-needed basis. Such deployments are based on the use of prebuilt virtual machine templates. Templates enable automated VM deployment and ensure virtual machines are configured in a way that adheres to the organization's established security policies.

These templates -- sometimes called images -- can pose a significant logistical challenge to private cloud administrators because server environments aren't static. There is always a security patch to apply, a new group policy setting to enable or an OS configuration setting to change. Because server operating systems tend to be fluid, VM templates quickly become outdated. As such, it is important to decide the circumstances that might warrant the creation of new templates, as well as who will be responsible for building them.

Some organizations build new templates on a monthly basis. Others only build new templates when there are significant new patches that should be included in the templates. There isn't a right or wrong approach to template maintenance, but you should establish a policy before your private cloud is put into production.

Defend against single points of failure

As you design your private cloud, be especially careful to avoid creating any single points of failure. For instance, Microsoft private clouds are based around the use of System Center Virtual Machine Manager (among other things). Because Virtual Machine Manager plays such a crucial role in the cloud, it is important to create a failover cluster that can make the Virtual Machine Manager server highly available.

Similarly, Virtual Machine Manager has a dependency on SQL Server. It does little good to make Virtual Machine Manager highly available if the SQL Server could become a single point of failure, so you should also make the SQL Server instance highly available.

Next Steps

How to plan a migration to a private cloud

Four steps to planning the perfect private cloud

Private cloud computing best practices

Dig Deeper on Cloud storage

Disaster Recovery
Data Backup
Data Center
and ESG