Essential Guide

Browse Sections


This content is part of the Essential Guide: Virtualization to cloud: Planning, executing a private cloud migration
Problem solve Get help with specific problems with your technologies, process and projects.

Private cloud computing planning from concept through deployment

Building a private cloud requires strict planning, but also flexibility. Changing technologies and attitudes will go a long way toward success.

Creating a blueprint on what your company expects to get out of cloud computing, getting buy-in from all departments, automating IT workloads, and transitioning to a self-service model are the initial steps in creating a private cloud. But the journey doesn't stop there.

Let's take a look at the five final steps, including initiating chargeback, bolstering security and monitoring cloud performance, to help you with planning so you can create a fully functioning private cloud in your enterprise.

1. Take a look at chargeback and showback models

As clouds form and workloads centralize, it is important for organizations to keep track of resource usage and verify that resources are consumed fairly and organizational priorities are accounted for.

A chargeback model is one of the most powerful yet most resisted forms of resource accounting. It can be difficult to implement chargeback in an organization with no history of accounting for resource consumption because it requires inventorying and justifying every server and application as it moves to the cloud. The process itself is good for an organization; it reduces waste, seriously curtails sprawl and puts pressure on application and system administrators to right-size virtual machines (VMs).

Moving forward carefully and working creatively with management and the CFO can yield some good solutions to budgetary issues. And care should be taken to make the chargeback process as unobtrusive and low-overhead as possible.

Organizations that cannot do chargeback right away can usually do showback, when reports are generated for management to show where resources in the cloud are being used. Showback is an excellent first step toward a real chargeback model and is useful in the initial stages of a private cloud to set budgets and expectations.

Many organizations that use showback techniques treat the model almost like chargeback. Specific projects and departments are assigned a dollar amount, except the bill is never sent to the customer. It is a powerful way to track and conserve resources, but the method can be completely foreign to developers, application administrators and other staff members who had never needed to justify or account for their resource use before. Care should be taken to ease staff into these new procedures.

2. Keep everything in its proper place in a private cloud

Security is always a big part of IT, and when you're moving toward the cloud, it is a good time to reconsider your approaches to cloud security. It's also a good time to consider new technologies.

One of the biggest changes an organization makes on the path to private cloud computing is internal cooperation.

While cloud computing doesn't necessarily require virtualization, the use of virtualization opens the door to features that include inter-VM firewalling and intrusion detection, agent-free antivirus scanning, and other features via APIs, such as VMware's VMsafe. While many clouds are built using traditional approaches to security, being open to new approaches can save time and money while adding flexibility. For example, inter-VM firewalling and intrusion detection may replace complex private VLAN setups, saving time and reducing complexity.

Another type of security measure is disaster recovery (DR), with its many products and options dedicated to maintaining off-site copies of VMs. Replication of storage at a VM level frees the storage administrators from having to acquire and maintain costly array-based replication licenses, WAN accelerators and Fibre Channel-to-IP converters. Replication can also be done to disparate arrays, which usually isn't possible with array-based options. You can easily manage recovery point objectives (RPOs) and recovery time objectives (RTOs) at a VM level with newer cloud-oriented options.

Some products also manage failover and failback and can significantly reduce the effort needed to maintain your organization's disaster recovery playbook by automatically applying DR rules to new VMs. Too often new servers are added to disaster recovery plans after implementation, leaving the servers unprotected in the interim.

3. Know that monitoring is crucial

Centralization of services into a private cloud has many benefits, but it doesn't make performance monitoring any easier. Relocating services often means an increased dependence on network performance, which, in turn, calls for extensive monitoring in addition to the tools that perform that task.

An increasing number of performance monitoring tools provide a single monitoring interface that is very useful to system, storage and network administrators who troubleshoot problems. Information gleaned from application monitoring system reports are just symptoms of a problem, not root causes. But it saves enormous amounts of time to be able to rapidly tell that what looks like a network problem is actually a storage issue.

Some performance monitoring tools also offer features that aid help desk and support efforts. Such tools are especially useful to detect intermittent problems and situations that do not trigger other performance alarms. In addition, the data can rapidly pinpoint the root cause of a problem.

Application monitoring is often greatly improved in a private cloud environment, mostly because of better documentation of requirements and the inventory process that organizations use to prepare for consolidation. Virtualization also provides high-availability and fault tolerance options at the virtual machine level, as well as high availability through the application within a VM.

4. Perform future-proofing within IT teams

Private clouds and virtualization technology decouple organizations from many problems that IT groups have been trying to solve for years. Centralizing, standardizing and automating workloads and workload management tasks frees time to do other things, such as keeping an eye on new technologies. That, in turn, reduces reliance on external consultants and builds knowledge and expertise in-house.

Computer scientist Alan Kay was on to something when he said, "The best way to predict the future is to invent it." That is absolutely true within organizations, too. IT teams need open minds about how organizational goals can be achieved. Ultimately, instead of just trying to keep up, IT staff will have more time to do things that move the organization forward.

5. Remember, we're all in this cloud computing thing together

One of the biggest changes an organization makes on the path to private cloud computing is internal cooperation. Years of building political and operational walls that divide parts of your organization serve only as a barrier to a cloud adoption project.

Private clouds can be quite expensive, and you will not realize any cost- or time-saving benefits when individual departments or divisions implement the technology on their own. Retaining flexibility and meeting the needs of all aspects of your organization are crucial as you centralize into a private cloud. To do this, though, all parties must be open and honest about their needs, have useful documentation and work in an iterative fashion.

Be sure to make room in a cloud plan for adjustment and change as everyone learns how to work in the new environment. Silos within IT need to disappear.

Very often, an organization's network, storage and system administrators work separately and become territorial about their work. The most effective implementations of virtualization and private clouds are supported by teams with members from each of these areas, working together for the benefit of the organization.

Applications in the cloud often depend on networking, especially when applications are centralized in data centers that are not local to the users. Storage is crucial to virtualization, and decisions made by storage administrators have long-lasting effects on service delivery, service-level agreements, costs and time. New technologies allow great efficiencies to be gained if IT staff members remember that it isn't their storage, their network or their systems. The cloud and its infrastructure belong to the organization.

Systems can be tuned to reduce load on networks and storage. Cloud environments have also begun to replicate in software what storage and network admins have always known as hardware features, such as firewalls and storage replication. The move to the cloud brings automation and standardization, which may cause hard feelings for staff members who are responsible for the way things are or whose jobs can be automated. Create good avenues of communication, assign no blame and make sure the IT staff understands that the changes will give them more important and more interesting work to do in the cloud.

The IT landscape has changed, your organization is changing with it, and experience with cloud computing continues to be a marketable skill.

Bob Plankers
is a virtualization and cloud architect at a major Midwestern university. He also contributes to and, and is the author of the The Lone Sysadmin blog.

Dig Deeper on Managing cloud infrastructure

Join the conversation

1 comment

Send me notifications when other members comment.

Please create a username to comment.

Indeed, helping IT realize the benefits of cloud in truly becoming a service organization is not trivial, yet the gains from providing service to the organization that meets (and beats) what they get from external cloud providers is well worth the investment. The vast majority of workloads under management by central IT are in direct support of development. As a result, it makes a great deal of sense to begin cloud efforts focused on servicing this group, (which often is responsible for the innovation that drives the business). This approach helps assure IT controls risk while providing development with building blocks such as common application platforms that speed application development and delivery, making central IT every business unit’s service organization of choice. 
This blog may explain further:
-- Derick Townsend, VP Product Marketing, ServiceMesh