IT Management

Steve Jobs’ June iCloud announcement (see: http://www.apple.com/icloud, http://www.usatoday.com/money/smallbusiness/columnist/abrams/2011-06-30-abdams-small-business-cloud_n.htm) and Microsoft’s prolific “To the Cloud” TV commercials have made “cloud computing” a household concept. However, there’s still much confusion about what cloud computing is and what it means to adopt it. While recent surveys show most IT organizations understand the mainframe should play a strategic role in their enterprise cloud strategy, they haven’t identified all the ways it can bolster competitive advantage.

What Is the Cloud?

With the excellent definitions provided by the National Institute of Standards and Technology (NIST) now published (see NIST special publication 800-146, May 2011, Lee Badger, Tim Grance, Robert Patt-Corner, Jeff Voas or visit http://csrc.nist.gov/publications/drafts/800-146/Draft-NIST-SP800-146.pdf), we should be able to get consensus on terms, delivery methods, and consumer/provider roles. The distinctions between private, public, and hybrid clouds, as well as the various delivery models and servers, should now be understood.

However, the NIST definitions don’t explain why cloud computing is so rapidly being adopted. Some enterprise IT professionals might still be in denial about it, but the truth is that enterprise IT should be asking, “When will we adopt a cloud strategy?” It’s not a matter of “should we,” or “why.”  

What’s the Urgency?

For small and new businesses, obtaining all IT services and infrastructure from the cloud is pretty much a no-brainer. The cloud is a business revolution providing low cost of entry for new innovations and much lower barriers for new competitors to enter your space. The way Netflix took on Blockbuster and now competes with the cable companies is an often-cited example. But cloud is relevant to existing, larger businesses, too.

Your IT department needs a cloud strategy because your company is probably already using the cloud. The wider definition of cloud includes things such as outsourcing. Line of business owners are increasingly signing up for Software as a Service (SaaS) directly, sometimes bypassing IT entirely in their decision-making. Even IT is bypassing IT by signing up for server images (i.e., Infrastructure as a Service [IaaS]) from Amazon rather than waiting for that extra test kit they need.

As the rules change, so must IT. Your IT department must take the lead to strategize on the new model and transform its role to more of a consultant/broker than simply a provider. For the forward-looking IT organization, this is an opportunity to recast itself from a cost center to a real source of competitive advantage, driving the discussion on right-hosting and exploiting existing resources.

Private Cloud for the Enterprise

As an enterprise IT organization plans for cloud, they will invariably decide on a mix of public, private, and hybrid solutions. Since most large enterprises rely on the mainframe for key business processing, they must start thinking about where the mainframe factors in the cloud.

What’s your vision of the enterprise cloud? Could you use Linux on System z to implement self-service development and testing on the mainframe? Does IBM’s new zEnterprise box, with its support of distributed operating systems, offer a realistic, perhaps cheaper and better, solution to VMware and vBlock? Though consultants tend to advise you to pick new workloads and servers, it’s more cost-effective and rational to start with your existing hardware and applications.

The mainframe has been providing many of the benefits of cloud computing for 40 years as matched up against the list of defining attributes published by NIST. As we cluster distributed machines into virtualized resource pools and tackle the problems of scalability, elasticity, multi-tenancy, and perhaps most important, security, we’re trying to solve issues the mainframe has already mastered. IBM has been baking these attributes into the mainframe operating system since the ’60s; most IT professionals recognize the mainframe as the most secure environment for critical data. Since security is a core issue as IT executives struggle with cloud, this advantage should put the mainframe on the front burner.

Since centralized computing is one way of achieving the economies that the cloud promises, IBM’s zEnterprise, which can run distributed operating systems, has real potential as a cloud platform. An enterprise IT strategy for private cloud must invariably involve multiple platforms, but when you focus on the strength of your mainframes to deliver on the cloud, it may simplify your strategy and minimize the server sprawl that has made data center management so challenging.

It’s the Business Service

To really leverage their mainframe for private cloud, enterprises need to be able to migrate or extend distributed applications to Linux for System z, or the new zEnterprise BladeCenter Extension (zBX), and connect them to transaction and database processing on z/OS. In the real world, enterprises have multi-tiered applications with data on the mainframe and front-ends on distributed machines, either physical or virtual. Translating this to cloud can provide great value.

A true enterprise private cloud will treat cross-platform applications as business services and let you manage the whole infrastructure as one: starting, stopping, scaling, backing up and migrating the service as a whole. Cloud computing isn’t as much about a new technology as a new model of delivering and consuming IT services. So, for traditional enterprises, though new workloads should be built the new way, the same principles could often be applied to existing systems and infrastructure. We need to understand how to use cloud delivery principles to renew and advance our management of existing services.

Consider a bank’s Automated Teller Machine (ATM) network. Many banks store key data on z/OS in DB2, which may be linked to other distributed databases, with the front-end on Intel-based servers. Because IT teams are often organized (and isolated) by platform, service-level management is challenging. A new enterprise cloud approach would see the entire workflow as a single service. By using a management tool to provision and control the service as a whole to apply a scaling policy across the platforms, you can ensure that, as front-end capacity and demand rise, the DB2 instance will be ready to process transactions at a matching rate.

Thinking on an Enterprise Scale

Strategically, your organization should be open to re-think everything, but initial cloud implementations would probably start with smaller applications to test viability. As you move to transform your most important applications, your mainframe strategy becomes critically important.

Generically, there are four main service architectures that involve the mainframe:

  • Linux on System z and distributed (e.g., Oracle on System z, front-end on Intel)
  • z/OS and Linux on System z (e.g., front-end/data on Linux on System z, key data on z/OS)
  • z/OS and Linux on System z and distributed (e.g., like the first architecture example, but data also on z/OS as in the ATM example)
  • z/OS and distributed.

While use of Linux on System z is growing rapidly, the z/OS and distributed architecture is probably still the most frequently found today at enterprise sites with mainframes. Many cases may be immediate candidates for migrating pieces to Linux on System z. In each case, the distributed portion could also run on the zBX blade server on the new zEnterprise-class machines, moving the entire service to run on a single secure box. Given the similarities between Linux across platforms, enterprise management would be simplified in this implementation. The key is to deliver and manage the service as a whole in a dynamic, scalable way. An advantage of the new zEnterprise is that all parts of the service on that box can be managed by a common provisioning policy.

It may seem that not all workloads are suitable for cloud, but that’s often more of a limitation of the underlying infrastructure and management paradigms. Fear that moving from dedicated physical servers to a virtualized environment will undermine resiliency and performance is really a comment on the maturity of distributed virtualization. If that same workload could be run on a mainframe, Service-Level Agreements (SLAs) would be easier to ensure. 

Conclusion 

We’ve entered a new era of IT cloud computing. Forward-looking IT organizations will seize this opportunity to transform themselves from being just a large cost center to being a vehicle for competitive advantage for their company. For traditional large enterprises with mainframes, IT will only deliver on the flexibility and agility that the new paradigm offers by building on the strengths of the past, embracing their mainframes, rather than trying to simply rip and replace. To be successful, they must focus on how they provide business services, embrace the cross-platform realities of business services, and find tools that leverage the strength of a heterogeneous environment that includes the mainframe, while truly reducing the complexities of provisioning and management. This is how you start building a real cloud strategy for enterprise IT.