How To Get To The Cloud

May 25, 2011
Step into virtualization first, and then from there make the leap to the cloud.

Cloud computing remains hazy to many business leaders in manufacturing. Just as its name suggests, the cloud computing model has different definitions, depending on who you ask.

Broadly speaking, we call cloud computing a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources, which can be rapidly provisioned and released with minimal effort or interaction. For financial decision makers, cloud computing is thought of as a paradigm shift from capital expenditure to operating expenditure.

For the manufacturing industry, cloud computing is viewed as a savior because it dramatically increases effectiveness and efficiency of business processes while decreasing cost (if applied correctly). While new materials and tools help manufacturers recover from the economic downturn, few technologies hold as much promise as the mainstream adoption of cloud computing.

Precursor to Clouds
Before we discuss clouds, however, we must discuss the process of virtualization because it is the prerequisite technology to cloud computing. In basic terms, virtualization is the creation of a virtual (rather than actual) version of something, such as a hardware platform, operating system, storage devices or network resources. This virtual machine acts like an actual computer. The general goal of virtualization is to centralize administrative tasks while improving scalability and workloads. Benefits yielded are power, space and cost savings across the board.

The obvious places to consider using virtualization are in manufacturing labs and data centers where one is trying to gain efficiency, reduce cost and save on physical space. Using virtualization, test engineers can gain efficiency by being able to spin up an array of diverse images to test products. Budget managers can gain space efficiency by getting more work done with a lot less hardware. Facilities personnel can conserve space by accommodating lots of workloads and running them on less hardware. IT administrators are delighted to be able to generalize virtual images on common shared hardware by creating a storage library of images and running them as needed.

Virtualization caught on initially as an efficiency tool. Then it was commercialized in manufacturing after it had fully harnessed its flexibility in the area of cloud computing. Virtualization crossed the chasm and went mainstream when the world outside the high-tech industry embraced the cloud as the next generation of computing for the forthcoming decade. Why is this so?

It is because all clouds implement some form of a specialty abstraction layer called the hypervisor, and this hypervisor is what provides the virtualization capability. Clouds are designed from ground up to function cohesively in 'on-premise' as well as 'off-premise' environments, and it displays four key characteristics. These characteristics include the ability:

  • to run software programs virtualized;
  • to scale the intelligent workload on demand;\
  • to support the notion of multi-tenancy for capacity sharing; and
  • to support a usage-based billing model for metering.

Virtualization gives the cloud the elasticity to scale on demand. It allows multiple operating systems to run on a common hardware simultaneously. Virtualization is also a prerequisite to intelligent workload management, which allows you to run application across physical, virtual and cloud environments seamlessly. Virtualization also serves as an enabler for multi-tenancy that allows customers (or tenants) to share infrastructure hardware, storage and bandwidth and thereby reduce the costs of computing. The usage-based billing model (if used) then allows subscribers to pay based on their consumption, which paves the road ahead for "computing as a utility" like electricity, gas and water in our houses.

Types of Clouds
Generally, think of clouds as datacenters in the sky, whose walls have disappeared and your burgeoning concrete data center has been deperimeterized. There are two schools of categorization for todays clouds. First is the deployment model, a cloud deployed as a service by vendors. Second is the management model, which describes the way a cloud is serviced or managed by your IT department. Looking through the lens of the deployment and management models, clouds are broken down further into three distinct categories in each:

  1. Deployment model:
    a. Private cloud
    b. Public cloud
    c. Hybrid cloud
  2. Management models:
    a. Software-as-a-Service
    b. Platform-as-a-Service
    c. Infrastructure-as-a-Service

Deployments of private vs. public vs. hybrid clouds have several differences from each other. Private clouds, which are typically stepping stones for manufacturing companies, tend to reside in the corporate data center, used by a single company, managed by IT, with a simple monthly chargeback model or allocation to the business units. On the other hand, public clouds reside in internet-connected data centers, are used by multiple clients, are managed by the service provider and are billed by consumption or usage. Lastly, there are hybrid clouds, which are similar to public clouds in the sense that that reside in internet-connected data centers and are used by multiple clients, but they differ in the way they are managed, which could be a combination of host, provider or IT, and the billing model which is a base monthly charge with excess usage billed by consumption.

The management models are how vendors frequently refer to clouds. Inside and outside of the manufacturing industry, the most common model of hosted computing is the Software-as-a-service (SaaS) model, under which vendors maintain their applications in a central data center and deliver their functionality to users over the internet. Manufacturing industry-specific providers like Plex, Apriso and are the leaders, while hundreds other are competing in this rapidly-growing market.

The second model is referred to as Platform-as-a-service (PaaS), and it is an IT development environment that allows programmers to write applications that will be hosted offsite. In this PaaS market, the most prominent players are (Salesforce.coms platform), Azzure from Microsoft, and Google App Engine from Google.

The third model is referred to as Infrastructure-as-a-service (IaaS), and it delivers raw computing power or data storage to the manufacturer from a data center. This on-demand computing power can be made to scale up or down dynamically, based on the users need. It is typically billed at an hourly rate for CPU usage and by the gigabyte for storage. The most significant player in this IaaS market is Amazon, who delivers computing power plus storage from its own data center via its Elastic Compute Cloud (EC2) and Simple Storage Service (S3) offerings.

In summary, cloud computing, coupled with virtualization, has changed the way manufacturers plan factories or develop goods in the industry. The cloud is expected to go mainstream within three years and get monetized over the next decade to offer the manufacturing industry a brave new world.

Dipto Chakravarty is the Vice President of engineering for Attachmate (formerly Novell). In addition to being a 24-year software industry veteran, Chakravarty is the author of two best-selling computer books from McGraw-Hill that have been translated in five languages. Chakravarty is also a member of the Cloud Security Alliance and is in charge of its Trusted Cloud Initiative.

Sponsored Recommendations

Voice your opinion!

To join the conversation, and become an exclusive member of IndustryWeek, create an account today!