LAS VEGAS -- The nascent world of software-defined tools can provide an avenue for enterprises to exploit the cloud -- once they figure out where to begin.
Software-defined anything, which ranges from software-defined data centers (SDDC) to a host of other applications, changes the way IT thinks about infrastructure and enables developers and users to access infrastructure in a way that is much easier to understand, said Donna Scott, vice president and distinguished analyst for Gartner, Inc., based in Stamford, Conn.
"If your business is pushing forward with more mobile apps, more Web apps, more analytics apps -- all which need to talk to your systems of record, your patient data, your financial data your product data -- if you are focused on that, the software-defined data center will get you faster times to deployments with your applications," Scott said during a session at the company's Data Center, Infrastructure and Operations Management Conference this week.
By 2020, Gartner estimates 70% of new applications will be deployed on cloud. IT is in the middle of a platform change and dynamism is the wave of the future, Scott said.
The last mile
Initial forays into private cloud often speed the time to get compute and storage from months to days or hours, but not to minutes because the process wasn't fully automated, Scott said.
Agility is in the eye of the beholder, and one of the last policy implementations to be fully automated is networking, she said. Manual steps such as connecting to an IP address or opening ports can slow the ability to connect with developers and end users. With the right policies and abstraction layers in place, SDDC can help enterprises innovate through greater agility and freedom from the underlying infrastructure.
Gartner analysts have taken to calling this spectrum of emerging services "software-defined anything," or "software-defined X." Beyond just SDDC, there's the broader notion of software-defined infrastructure, which includes data collected in factories, homes, automobiles and other areas outside the traditional data center, as well as software-defined application services
Software-defined networking is an even less adopted technology, with only hundreds of implementations, Scott said. But it does have a potential future in several areas, including disaster recovery, especially if policies can be set across data centers.
Bob Ess, vice president of enterprise IT infrastructure for MedAssets, Inc., based in Plano, Texas, attended the session to learn about the software-defined technologies, but said it would probably take "18 months and more resources than I've got" to implement any of them right now.
"The technology is very interesting," Ess said. "I just don't know if the software vendors have mature products to go into production right now."
Despite his interest, Ess said he still has concerns about the change to SDDC and how that would impact existing workloads.
Another concern is the schedule for when these various technologies will come to fruition, said Greg Anson, enterprise infrastructure architect for Unum Group, based in Chattanooga, Tenn. Analysts expect the various components of software-defined technology to reach maturity at different times over the next decade, so at this point it might be more about creating a mindset to get ready for that time, he added.
"How do I get to a software-defined data center when they all have different timings on when they're expected to be delivered, and what am I supposed to do today?" asked Anson.
Anson said he would like to do more with these emerging technologies and cloud, but it's a matter of proving the value in investing in such a major shift.
"It comes back to total cost of ownership," Anson said.
The two central ways to exploit software-define data centers in the cloud is through the cloud management layer or through cloud brokerage vendors, Scott said.
"Software-defined is about reusability and creating those blocks of standard infrastructure that you want to offer up -- a standard menu of services that you can offer to your developers or your users," Scott said.
Where to begin with SDDC
A good place to start with SDDC is through faster provisioning, which is the primary driver for most uses today, Scott said. DevOps is another key way to take advantage of SDDC.
The first step in using the technology for DevOps is to create a one-step automation build to test and deploy, Scott said. After that, develop automated rollback to a known 'good' state, then build a model of interdependencies and develop test iteratively.
Once that's done, the first production deployments can begin with complete end-to-end infrastructure and application, and finally, additional releases may be completed or incrementally upgraded.
It's important to start small and be mindful of the uses before delving into SDDC, but it's also critical to keep an eye on changes in this emerging space, Scott said. Operations managers will have to develop new skill sets to continually monitor compatibility because cloud vendors change APIs and it will be up to the user to stay on top of these changes and adjust accordingly, she added.
"It's important to understand how your orchestration is using the various APIs," Scott said. "How those APIs change over time is a critical part of the infrastructure operations."