The Software-Defined Data Center: 5 Tips For Success

Moving to software-defined infrastructure can help companies be more efficient and agile, but requires reengineering of IT processes.

Aaron Bawcom

April 10, 2015

6 Min Read
Network Computing logo

A primary goal of the software-defined data center is to help IT organizations be more nimble and deliver rapid, cloud-like services to users. There’s also the promise of saving money by delivering IT in a more coordinated, streamlined fashion. Infrastructure components and services are consolidated, fully automated, driven by business policies, and centrally managed for performance. A SDDC can monitor demand and respond automatically in seconds by spinning up appropriate resources.

Vendors that develop software-defined solutions for storage, networking and compute predict heady results, such as 75% capex savings and 55% Opex savings. Technologies that support the software-defined data center  can help eliminate the traditional data center silos of compute, networking and storage, and build upon the server virtualization infrastructure that has matured in many midsize and large companies.

Yet technology alone, as always, can’t deliver the goods. SDDC requires some rather robust reengineering of IT processes to bring the cost savings, productivity gains and business agility it purports to deliver. Here are  five strategies to maximize your success in moving toward software-defined infrastructure.

1. Start small

The greatest barrier for companies in adopting a software-defined data center is inertia. There’s been a misconception, partly due to vendors, that companies must dive in headfirst and transform the entire data center operations simultaneously -- and/or commit to projects that take on too much, too soon. This is intimidating and not necessary.

Instead, begin your SDDC journey with a single, small-scale project that’s associated with a specific low-profile service or activity addressing one plane of the SDDC environment --compute, networking, or storage. For instance, moving a storage workload involving a database service is a perfect project to get your feet wet. Using a technology like VMWare Live Migration to dynamically move VMs without disrupting service is how a company can learn about SDDC and achieve quantifiable benefits, such as 100% uptime.

Conversely, targeting the e-commerce site for the first foray into SDDC could invite disaster. This type of project involves multiple application services (inventory, order management, shipping) and as a result, requires that solutions for software-defined storage, networking and compute all work together seamlessly. If there is a failure or delay in the new infrastructure, which is likely when trying something new, the result is alarmingly visible to senior management. CEOs don’t like it when revenue generating systems go down, even for a minute.

By starting with a small-scale, non-mission critical project, IT can learn quickly and refine processes for the next project and therefore build SDDC expertise without incurring business risk.

2. Have the necessary skills

In the software-defined data center, the IT team needs people who understand systems automation and orchestration. These skills span distinct technology areas and are typically found in individuals who work closely with the business, with external service providers, or who have had cross-departmental roles such as a systems administrator.

Secondly, understand that SDDC technologies are usually highly vendor specific. Selecting a Cisco solution, for instance, will require people with expertise in Cisco networking. It’s far easier and less risky to choose platforms based on the skills you have in-house, instead of trying to retrofit people toward an unfamiliar technology. Even so, despite having a highly capable team, SDDC is early-stage enough that IT organizations still need to spend money on training, support and outside consulting.

3. Evaluate legacy technologies and vendor contracts

No intelligent IT leader wants to (or can afford to) rip and replace all systems and vendor relationships in order to adopt SDDC capabilities. IT should factor in business priorities for vendors based on long-term contracts and purchasing power, and then align SDDC purchases with those relationships.

A company that has standardized on Citrix desktops instead of VMware technologies did so for reasons such as cost, capability, and vendor sourcing agreements. For that organization, switching to a different vendor may overcomplicate the sourcing strategy, making it less likely to choose SDDC solutions from VMware. There’s also a need to evaluate hardware end-of-life status. If IT wants to deploy Cisco SDDC but has HP networking infrastructure that’s only two years old, it makes better economic sense to choose HP.

4. Rethink the IT organization

The mantra of “no more silos” has reached the end of the road with software-defined infrastructure. There’s simply no way that the CIO can continue to run IT as in the past, with separate groups for storage, networking, servers and applications all working happily yet apart. Maintaining silos is a technology barrier for SDDC, where data must flow freely and be coordinated at a high level. 

Silos are also counteractive to the end goal. In a perfect world, SDDC provides more relevant information from all the components, and then distributes that data continuously across IT to automate better decision-making and management. It’s not about server uptime or storage efficiency, but can the user complete a transaction or task within the required service level? How quickly can IT upgrade systems? Does the application work in ways that support the business unit’s goals?

Roles also change. If a company deploys a software-defined compute product such as VMware, that will affect the storage or network group. Those individuals will need to optimize and deliver services based on a virtualized infrastructure and upon standards that are optimized for SDDC. Above all, people must collaborate in new ways, and frequently, across the boundaries of old.

5. Use metrics for the business

IT departments have been monitoring performance in manual and automated ways for many years, yet those metrics often don’t mean much to the business. If 10% of the servers are 30% utilized, there’s no (seemingly) clear business benefit from increasing utilization. Instead, select a handful of metrics that will make a difference to business managers and in defining success for the new infrastructure. Sometimes, those metrics will change from project to project.

Regardless, metrics should tie back directly to a business benefit. Choose metrics that can demonstrate how much more agile, efficient, and effective you can be supporting users in the new SDDC-oriented environment. Common across all initiatives is speed of deployment. Agility is the prime driver for IT today and software-defined infrastructure should be able to cut the time to deploy, upgrade or fix a user service dramatically, even by 90%.

Other metrics include:

  • Speed to adapt to real-time demand, by measuring response times to and from servers and clients

  • Ability to move storage resources with zero downtime

  • Speed of repurposing infrastructure to host different applications

  • Speed of diagnosing problems

  • Ability to scale infrastructure without user downtime

  • Speed of replacing systems

  • Speed of moving workloads to alternative infrastructure, such as from AWS to VMware

  • User satisfaction

  • Ease of use

  • Total costs

There’s been some debate as to whether SDDC is a technology or a methodology. In essence, it’s both. But if the IT organization begins the discussion with arguments around which vendors to use, the big picture of SDDC’s value could get lost or undermined. SDDC requires a new orientation to delivering and managing IT, one that’s founded on business prioritization, collaboration and speed.

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights