IT Management

Cloud computing represents a disruptive paradigm shift that will eventually shake most IT organizations to their roots. New platforms, new applications, and a new generation of tech-savvy business users will put enormous pressure on IT, requiring them to weave together the best of traditional IT with the best of what’s available in the cloud to deliver business value to their organizations.

There’s a striking parallel between IT’s transition to cloud computing and manufacturing organizations migrating from a factory model to a supply chain model. In the past, manufacturers typically ran closed, standalone factories—raw materials in, finished products out. Now, successful manufacturers leverage a global supply chain, selecting components from a variety of sources according to cost, availability, and quality requirements. The base components remain important, but the organization ultimately creates value by building products using a diverse mix of sources, both internal and external. Systematic supply chain management techniques improve agility and quality while reducing cost, translating into competitive advantage.  

Today, most IT organizations also work according to a factory model—assembling, deploying and managing most IT services in their own data centers. In contrast, cloud computing is a fresh new approach—one that’s much more about running a dynamic IT supply chain. The transition from IT as factory to IT as supply chain represents a new, different way to think about IT’s role; it promises substantially increased efficiency and agility, just as manufacturers found with supply chains.

Although the journey to cloud computing might sound a long way off, business users are already making the first moves toward building an IT supply chain—sometimes even outside of the IT department’s control. Sophisticated business users understand there are alternatives beyond the IT factory, and they’re increasingly selecting cloud-based solutions rather than traditional ones. Modern cloud applications usually offer adequate capabilities, even though IT might express concern about certain issues such as security. They’re often priced so low as to skip past traditional corporate procurement processes. And, being cloud-based, there’s also an important sense of instant gratification where the user can be up and running almost instantly. These alternatives lead some business users to envision a future where the complexity and cost of running a large data center becomes someone else’s problem. In this way, they hope to reap drastic benefits in agility, quality, and cost.

New and emerging organizations are also responsible for accelerating the trend toward cloud computing. Today, few start-ups build an internal data center. Rather, they turn to cloud email, financial and human resource applications, relying on non-traditional partners such as so-called Managed Service Providers (MSPs). Over time, the increased agility and lower cost of cloud approaches become the weapons emerging organizations wield against more established competitors. Today’s organizations must adapt, or risk increasing threats from a new, formidable breed of competitor.

Faced with these realities, IT organizations frequently respond by attempting to act more cloud-like, promising internal business users the same flexibility and cost advantages found in the cloud on their internal systems. Larger organizations often have adequate scale to compete against many external cloud providers. IT also understands that although cloud services are often extremely attractive from a cost or agility perspective, they’re often suspect in terms of quality of service, security, and risk. These factors create an opportunity for various internal, or private cloud initiatives.

Cloud computing also fosters a tendency toward a sober rationalization of IT capabilities vs. business value. Cloud computing offers many new possibilities, enabling IT to choose the right quality of service for the job. For example, not all services require “five nines” availability; sometimes “good enough” is truly good enough. This thinking permits the organization to mix and match, using potentially more costly internal capacity for the most critical services, while adopting cloud for less important applications. Unlike prior generations of outsourcing, cloud is infinitely flexible this way; you only outsource selected services where there’s some gain to be had, not the whole operation.

The availability of new and good enough cloud capabilities, combined with pressure both from inside and outside IT, creates a sense of urgency to discover an optimum balance of internal and cloud services. Doing so allows IT to address the organization’s needs, while simultaneously seeking out cost-efficiency, agility, and quality advantages. In effect, CIOs are balancing the often-conflicting needs of the organization by juggling suppliers—becoming supply chain managers in the process.  

Meanwhile, the mainframe fills a unique place in large organizations, and IBM’s new zEnterprise System takes the mainframe’s traditional strengths even further.

Most textbooks on supply chain management would explain that a key requirement is having a disciplined, systematic way to understand and monitor supply chain components. Immature practitioners focus exclusively on cost metrics, but learn that a more balanced approach is necessary, and metrics such as quality, risk and other capabilities are quickly included. On any matrix including risk and quality of service, mainframes excel.

Consider also that few cloud providers are likely to want to tackle the extreme quality of service attributes associated with the most strategic mainframe applications. Applications that exploit the mainframe’s unique high availability and quality of service features (such as Sysplex, advanced security, etc.) will prove challenging to re-host using public cloud alternatives. Although this will likely change over time, today’s sweet spot in the cloud is all about high-value and rapidly deployed capabilities with good enough qualities of service. Again, this speaks well for ongoing mainframe participation in cloud initiatives.

Also, even sites with strong commitments to cloud computing are hesitant to surrender too much control, especially to emerging, unproven cloud providers. In these cases, organizations typically lean toward hybrid approaches where some capabilities remain traditionally hosted, while others move to the cloud. While the largest cloud providers achieve mainframe-class qualities of service through enormous scale and sophistication, hybrid solutions are often much more fragile on the client side of the equation. It does little good to have all the redundancy and scale of a cloud provider’s enormous data center if critical data remains on-premise, stored behind unhardened servers. What’s needed is a self-contained, on-premise system that can be as resilient, scalable, and robust as the more massive infrastructure on the other end of the pipe—and the mainframe is ideal in this role. 

As organizations increasingly turn to private cloud and hybrid models, the zEnterprise system makes an attractive cloud in a box platform, thanks to these specific capabilities:

  • Flexible capacity. Cloud applications tend to be variable and dynamic in terms of capacity requirements. The zEnterprise system is based on ultra-fast 5.2 GHz processor cores, giving it unmatched capacity to host large workloads. Also, the system is dynamically configurable, enabling customers to add (or remove) capacity as needed to meet workload demands. Unlike competing platforms, zEnterprise also allows for dynamic, non-disruptive upgrades to devices and channels, network paths, memory and so forth—not just processor capacity.
  • Workload diversity. More than any other system, zEnterprise excels at running mixed workloads, no matter how diverse. While it’s completely at home running traditional mainframe workloads, it’s equally effective running large-scale Linux deployments using z/VM virtualization technology. The zEnterprise BladeCenter Extension (zBX) will give mainframe users a simple way to connect mainframe and Intel-POWER7-based servers in a single logical server—truly a data center in a box. Unlike other private cloud approaches, this permits sites to run complex, multi-tier applications side-by-side in a single system. Not just a convenience factor, this approach also has a variety of performance, reliability, and security advantages.
  • Energy and cost efficiency. For a given workload, zEnterprise can be more than 90 percent energy efficient than alternative platforms, translating into direct and substantial savings over the life of the system. Likewise, because it’s always easier to manage and secure a few large systems than hundreds of small ones, staff and system administration costs are typically much less on zEnterprise. With impressive reliability and availability, zEnterprise can usually run the most demanding applications at much lower cost than many cloud alternatives.

Up until now, the missing part of the puzzle has been a cloud platform capable of spanning mainframe and other systems, whether internal or external. Part of the challenge has been that emerging cloud platforms tend to deliver applications in the form of virtual machine images, and this isn’t a particularly useful construct for mainframes. Sites prefer to have a more granular approach that might let you deploy different application and middleware components on the optimum architecture, depending on business priorities, resource availability, and quality of service needs.

Consider this example. A site might have an application that requires a Web server, an application server and a database, but if packaged into a single virtual machine, deployment flexibility is limited. While the application could be deployed as a monolith to an internal cloud or to an external service such as Amazon EC2, the optimal solution might be different. Perhaps the Web server is best deployed on commodity Intel platforms, the application server on zEnterprise WebSphere, and the database on IBM DB2 for z/OS. The optimal deployment could change seasonally, depending on business conditions. A virtual machine-centric packaging approach makes this sort of flexibility difficult. Returning to the supply chain metaphor, it would be like an automotive factory only delivering complete cars, rather than enabling different factories optimized for engines, bodies, and electronics.

Software vendors are seeking to permit customers to leverage the full power of all platforms for private clouds—internal, external, and mainframes. The more sophisticated emerging products offer fine-grained control of complex applications in a cloud platform. Customers can visually compose applications by selecting components from a catalog of available services that are completely abstracted from the underlying hypervisor or platform. This level of abstraction makes it easy for any application developer or architect to create a virtual data center in a self-service way. It then becomes just as easy to work with internal private clouds as with external providers. Although limited to Intel architectures today, there are plans to extend these concepts to the mainframe in the future.  

A high percentage of the world’s most important commercial data resides on the mainframe, making it imperative for new generations of cloud-deployed applications to accommodate the mainframe. zEnterprise and emerging cloud approaches promise to change the mainframe perception dramatically, propelling it into a new role as the platform of choice for future private cloud deployments.