SYS-CON MEDIA Authors: Doug Masi, Mat Mathews, PR.com Newswire, David Smith, Tim Crawford

Related Topics: Cloud Expo, Java, SOA & WOA, Virtualization, Web 2.0, Security

Cloud Expo: Article

Goldie Locks and the Three Clouds: The Rise of the Enterprise Cloud

A cloud needs to be more than an infrastructure dispenser – providing small/medium/large chunks of infrastructure for each user

We all know the story of Goldilocks and the three bears, but have you heard the one about Goldie Locks and the three clouds? This tale is playing out throughout the IT marketplace.

Goldie Locks - an IT executive for a state government - has once again found herself in a dilemma. "If only I could choose one of the three options," she sighs. Goldie's dilemma is a result of competing requirements within her enterprise. Regarding infrastructure costs, Goldie has been told to "do more with less."

"If someone says that one more time, they're going to have porridge thrown at them," she huffs. Goldie knows that standardizing infrastructure requirements to serve the business and its processes securely, reliably and quickly is a proven way to reduce capital and operational costs. On the other hand, various business units and their departments have specific requirements for their mission-critical applications. They are resisting giving up control.

While many of her colleagues suggest she use the public cloud, Goldie believes that the security implications would be a deterrent to acceptance within her enterprise. Though she has done a thorough job investigating various cloud computing models, she needs to put together a request for proposal (RFP) to start searching for outside help with her dilemma.

She begins by taking into consideration the three standard deployment models of cloud infrastructure and their hybrid combinations - as defined by the U.S. National Institute of Standards and Technology (NIST) - and determining whether these are a fit for her enterprise:

  • Private cloud: Provisioned for exclusive use by a single organization comprising multiple consumers (e.g., business units).
  • Public cloud: Provisioned for open use by the general public.
  • Community cloud: Provisioned for exclusive use by a specific community of consumers from organizations that have shared concerns (e.g., mission, security requirements, policy, and compliance considerations)
  • Hybrid cloud: A composition of two or more distinct cloud infrastructures (private, community, or public) that remain unique entities, but are bound together by standardized or proprietary technology that enables data and application portability (e.g., cloud bursting for load balancing between clouds)

"Death by Committee" is her thought after analyzing these models. Each is too narrow for her enterprise. These cloud deployment models only provide the "how" without truly understanding the "what" and "why" of her situation. The IT side of the decision is obvious - drive down operational and capital costs to give the IT team time and money to solve strategic issues for the business. The best way to do this is to standardize processes, automate tasks and share infrastructure and administrative resources as much as possible. This is, in general, what a cloud provides.

But IT is serving a set of key stakeholders who have requirements beyond infrastructure. These stakeholders are the application/business owners who rely on IT to support delivery of their revenue-generating services and products. They are not well served with the three deployment models defined above, nor are they served by a hybrid of those models.

A Look into Goldie's Enterprise
In general, an enterprise consists of distinct parts (such as business units) that serve different customers, have different financial results and offer different products and services. They are fairly autonomous, but all operate from a common set of financial resources and processes, a common strategy and common metrics that determine success.

Consider Goldie's enterprise, which is a consolidation of a number of state agencies:

Her enterprise is partitioned into two high level branches (State Police and Transportation), where each branch consists of multiple, semi-autonomous departments. Each of those departments is interested in controlling their own infrastructure. Furthermore, their security and administrative processes differ from one department to another. Different departments must comply with their own levels of privacy, availability and service metrics.

For example, the Department of Public Info Office, within the State Police branch, may require highly predictable, millisecond response times for public users. In this instance, it may make sense to use public cloud infrastructure for the web servers, since there may be a requirement to scale up very quickly to reach high workload demands.

The Department of the Deputy Commissioner, also within the State Police branch, may also require specific infrastructure services, processes, automation and regulations, such as "hardened" OS images and encryption for all transmission of information.

The Department of Highway Administration, within the Transportation branch, must guarantee that their web site is available 99.999% of the time. They may have high availability requirements that demand duplicate resources at a disaster recovery site, as well as requirements for high-availability configurations.

NIST Deployment Models and the Enterprise - Square Peg, Round Hole
Goldie's enterprise cloud must be structured to support these multiple "parts." In turn, these parts can themselves have parts, and so on. This is similar to many of today's enterprises, which are the result of consolidating other businesses and agencies that need to function in a semi-autonomous fashion, but are still members of the larger organization.

A cloud needs to be more than an infrastructure dispenser - providing small/medium/large chunks of infrastructure for each user, without considering the unique requirements for different groups of users. Goldie knows that today's cloud products and services do not meet the needs of her enterprise stakeholders. She would like to deploy a single, centralized enterprise cloud that allows business units and their sub-units to:

  • Share underlying virtual resources as one large collection of cloud resources
  • Allow end users, such as developers, testers, demonstrators and system admins, to use a simple service catalog to manage the lifecycle of all cloud resources in the same manner
  • Set up autonomous administration, with unique policies and processes, as required
  • Allow business units to deploy their entire spectrum of applications, with unique service level objectives for development, test, production, mission-critical and regulated workloads

Now, let's see why the current cloud models cannot address these requirements.

Private Clouds
Goldie has looked at all of the currently available private cloud products. "These are too small-minded," she thinks. Every private cloud offers "multi-tenancy," which allows each business unit to manage its allotted set of cloud resources. But none of them offers any additional structure beneath the first level. Many of Goldie's business units have their own autonomous sub-units that require unique policies, processes and resources. They will want their own cloud, which does not meet her first requirement.

Public Clouds
She then turns to the available public clouds. "They are big and cheap, but my stakeholders do not want to expose their mission-critical or regulated applications." She chuckles thinking about a specific security dink she knows who actually turned pale when she suggested a public cloud for his application. On the other hand, she is painfully aware of some development teams that are slipping under the radar and deploying virtual resources in a public cloud for test and development. It's cheap and cheerful, but it's not handled by the centralized IT department and it exposes the business to risks.

Community Clouds
A community cloud offers cloud resources to a like-minded set of users / administrators. These users have agency-specific requirements, such as service levels, privacy, etc. If individual community clouds are deployed, then Goldie cannot optimize the sharing of all of the cloud resources. "This just isn't right at all," she says.

Hybrid Clouds
The final NIST deployment model does not provide any capabilities over and above the first three models. Instead, it is defined as one or more distinct instantiations of either a private, public or community cloud. Goldie has looked at all of the hybrid cloud management services and products, compared them to her requirements and decided that it doesn't meet her needs.

The Rise of an Enterprise Cloud
Through her analysis of the traditional cloud models, Goldie concludes that none of them are quite right. What she's looking for is a cloud that can address requirements unique to her enterprise. Let's refer to this as an "Enterprise Cloud." An Enterprise Cloud provides the capabilities of private, public and community clouds within a single cloud management platform that can support heterogeneous processes and requirements.

Goldie eventually conceived of the Enterprise Cloud illustrated below. It consists of a blend of internal datacenter resources, as well as resources provided by one or more public clouds. These are the "raw ingredients" that are abstracted into "cloud resources." Each agency can choose the specific cloud resources it needs to meet its requirements, including high availability, speed of deployment, cost, compliance with regulations and low latency response times.

Cloud-wide administrators, as well as specific agency and sub-agency administrators, are responsible for managing cloud resources through one "single pane of glass" interface. Aside from the properties of the cloud resources, their life cycles are all managed in the same manner, independent of where the raw materials came from. The end users of the cloud (e.g., testers, developers, infrastructure administrators) can be isolated from the underlying source of the raw resources. For example, an application could use public cloud for its web-facing tier, a low-cost set of internal cloud resources for its application tier and a highly regulated, encrypted and hardened set of cloud resources for its data layer. Goldie thinks of this as a "Hybrid Enterprise Application."

Goldie concludes that she needs to strike out on her own and develop a unique RFP that reflects her mental image of an Enterprise Cloud. If she settles for the types of clouds that are enumerated in the NIST document, she will never convince the various stakeholders to share a single cloud.

By focusing on key requirements, such as a single management framework across the enterprise, using public clouds and the datacenter to store virtual resources and providing a hierarchical multi-level tenancy structure, Goldie decides that she has finally found an Enterprise Cloud that is "juuuuuust riiiight."

More Stories By Michael A. Salsburg

Dr. Michael Salsburg is a Distinguished Engineer and Chief Cloud Solutions Architect for Unisys Corporation. He holds two international patents in infrastructure performance modeling algorithms and software. In addition, he has published more than 60 papers and has lectured worldwide on real-time infrastructure, cloud computing and infrastructure optimization. In 2010, Dr. Salsburg received the A. A. Michelson Award from the Computer Measurement Group – its highest award for lifetime achievement.

Comments (0)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.