A Data Science Central Community
The mainframe’s strongest assets – scalability, reliability, availability, security and manageability – are the very characteristics that enterprises are most concerned about as they consider moving business-critical services to a cloud-based architecture. Can the mainframe be a cloud platform?
Recent surveys underscore customer confidence in the mainframe’s ability to perform in new service delivery models. Growing numbers are considering IBM’s zSeries Enterprise as a cloud computing platform. As a result, mainframes, which have been central to enterprise processing for decades, will continue to be relevant as they find new life in modern computing architectures.
With the introduction of the new entry-level BC12 (Business Class), IBM is extending use cases to cloud and big data analytics. If one defines cloud as a resource that can be dynamically provisioned – allocated and de-allocated on demand – and made available with good security and management controls, then all of that functionality already exists on the mainframe. IBM is also a contributor to OpenStack, adding z/VM hypervisor and z/VM operating system APIs to the community.
Some skeptics cite the lack of complete self-provisioning on the mainframe as a hindrance that also reinforces the old-style command and control by IT. However, newer mainframe software tools do enable self-provisioning. It is also worth noting that in any cloud architecture, including those built on x86 servers, users would still need to obtain permission for services. As cloud computing matures and mainframes offer even more computing power at lower TCO, more organizations will consider the platform for cloud initiatives.
As for licensing costs, the zSeries has a little-used “on-off” feature that lets mainframe administrators turn a processor core on for a limited time only, thereby paying short-term rates for IBM software, rather than purchasing an annual license based on the number of processor cores. Improvements to Websphere and the z/VM operating system, coupled with more aggressive pricing by IBM, gives customers all the tools for a self-provisioned, mainframe-based cloud. As the mainframe runs more of the same software as other high-end servers and gains expanded service management capabilities, customers will increasingly view it as a workhorse at the high-end of the server market rather than as expensive legacy infrastructure.
IBM’s architecture extends use cases, draws new customers
IBM’s new software architecture was designed with extensive input from large customers who wanted to extend the fortress-like capabilities of the mainframe. It allows multiple workloads on mainframe, Power 7 and the System x servers to be centrally managed as a single, virtualized pool of resources. The Unified Resource Manager software extends to other systems integrated in the new architecture.
Additional capabilities, such as the DB2 Analytics Accelerator speeds workloads on Cognos and SPSS analytics. The BC12 also attaches to Netezza and newer PureData for high-performance data warehousing.
IBM set the stage a decade ago when it made Linux native to all of its mainframes. While mainframes have been using virtualization technology for over 30 years, since the introduction of the z/VM Virtual Machine operating system, the integration of Linux meant that enterprises could run virtual x86 servers on a mainframe. With the ability to combine mainframe and distributed computing platforms under a common management umbrella, organizations can consolidate and virtualize on a single architecture at the heart of which is the mainframe.
Interestingly, demand has been strong in emerging markets, including BRIC countries and other parts of Asia and Latin America. Rapidly growing customers in global markets are using the zEnterprise and new management platform to scale their business. Customers that built data centers on different platforms, including Linux, are using the zEnterprise to consolidate onto a single platform that natively runs Linux.
Also, as existing and new customers put more workloads on zEnterprise, it can become a platform from which to deliver private and public cloud services. IBM has added software to accelerate complex analytic workloads more efficiently and additional blades for BladeCenter Extension (xBx).
Energy consumption tilts the cost analysis
Many enterprises that are building out cloud computing platforms have gravitated toward x86-based distributed architectures due to the high cost of mainframe hardware and licensing costs and the shortage of IT professionals with mainframe skills. As enterprises increasingly model cost of service against usage to include electricity consumption, cooling, building/facilities costs and required manpower to maintain systems uptime, the granular reporting of mainframe resource usage compares favorably to standards-based x86 servers. Past assessments of total cost of ownership (TCO) that only covered systems and software acquisition and maintenance costs, which were then averaged across the user base, are giving way to a broader set of metrics that include the sizeable contribution of energy costs to opex.
The mainframe is likely to do very well when its power / performance and scalable management are compared to industry standard systems. This is partly a consequence of the platform’s architecture and design, but also down to the fact that mainframes typically run consistently at utilization levels higher than many other platforms can reach for any sustained period of time.
The pressure to model cost of service against usage is certain to increase as organizations seek to make the most of their IT infrastructure by creating highly responsive resource pools (“private cloud” or “dynamic infrastructure”) to minimize IT costs while maximizing business value. When looking at centralized and consolidated infrastructures the question now is whether the mainframe is worthy of greater consideration than it currently receives, both if the organization already has such systems in place but also perhaps as a new investment.
It is clear that getting the skills and tools in place to implement dynamic IT will be a challenge whichever route is taken, and contrary to common perception may even justify the investment in mainframe technology if the organization does not currently use it.