August 28, 2013 | Written by: Paul DiMarzio
Share this post:
In the first part of this series, I interviewed Jeff Frey, an IBM Fellow and Chief Technology Officer of the System z platform, about the software-defined environment (SDE), what it is and how it can help. In this second part, we dive deeper into the SDE conversation.
How does SDE relate to cloud computing?
I have some background in cloud computing, and at this point in the conversation I was thinking that SDE was just another way to express the cloud concept of infrastructure as a service (IaaS). So I asked if the two were one and the same.
Jeff says the difference is subtle but important, and he gave a great example: “Suppose I wanted to host my new web business on Amazon’s EC2. The interfaces that Amazon exposes to me that let me choose what I need, give me access, create my account and manage my billing—that’s IaaS. Under the covers, how Amazon actually provides the resources to me that fulfill our contract—this is where SDE would help.”
In other words, SDE is the engineering and management tooling that makes it easier for an infrastructure provider to organize their resources and enable operational efficiency; IaaS builds the business sense on top of this dynamic infrastructure.
Jeff was also quick to point out that SDE is not at all limited to cloud environments. Any IT shop that wanted to improve its business responsiveness would benefit from SDE.
What’s the role of the mainframe in SDE?
I work for the System z brand, so of course everything I do—and every blog I write—needs to tie back to IBM’s flagship zEnterprise system. So how does SDE relate to the mainframe?
The goal of SDE is to make IT infrastructure dynamic and responsive. For this to happen, the servers, hypervisors and virtualization managers that make up that infrastructure need to be able to exhibit certain characteristics, such as: multitenancy, rapid resource provisioning, elastic scaling, policy-driven resource management, shared infrastructure, instrumentation, accounting and audit trails and so on—and express these capabilities through application programming interfaces (APIs) that enable high degrees of dynamic and intelligent automation. If these characteristics don’t exist, they will have to be built in or added on.
The modern virtualized mainframe stack (zEnterprise and z/VM with SMAPI, for example) already exhibits many of these characteristics. All that’s needed is OpenStack integration for System z to be the cornerstone of any SDE.
There are no limits!
SDE addresses the needs of IT organizations, allowing them to manage their compute, storage and network assets programmatically to simplify the creation of services that drive down on this virtualized infrastructure. Many enterprises—especially small and medium-sized firms—would like this concept of programmatic automation to be enabled further up the stack, into the middleware and software service layers.
IBM has lots of experience in this area, most recently exhibited in the PureApplication System line. In this recent eWeek article, IBM Distinguished Engineer Matt Hogstrom—CTO for IBM’s software-defined environments efforts—talked about how IBM intends to focus on delivering turnkey, application-ready solutions that deploy software as a service (SaaS) on top of an SDE.
There really is no limit to the automation of the data center that is built on SDE technologies. A software-defined environment can help IT shops bring order to chaos, allowing virtualized IT resources to be dynamically and programmatically assigned based on the characteristics of the workload requesting resources. Twenty-first century applications running on twenty-first century infrastructure!
What do you think—is SDE an imminent reality or marketing hype? Will it help you deliver better value to your business? Add your comments below or engage me on Twitter, @PaulD360.