Michael Heil remembers when network convergence began to make sense to him. A few years ago, the manager of infrastructure engineering for Cone Health, a regional health care provider in central North Carolina, was asked to stand up a large virtualized SQL environment running Microsoft Amalga in 90 days with no additional resources. About the same time, he was also asked to implement a new document management system — again, with no extra hands on deck.
Heil knew that Cone was just beginning a growth spurt, and he knew that despite having a virtualized and consolidated data center, the company was already starting to hit the limit on rack space, power, and cooling. “We had to figure out how to do this with what we had on hand,” he says. “That’s what drove us to adopt a Cisco Unified Computing System platform.” By converging its storage and data networks, Cone was able to achieve some impressive gains in efficiency while cutting costs, says Heil.
“In the past, standing up a new VMware server would have required three different people three or four days,” says Heil. “My server guy would have to arrange time with a network tech and a storage person, so they could do everything in the proper order. After we implemented Cisco UCS, one person could deploy an ESX server in three or four hours.”
What’s not to like? The problem is that as enterprises converge their storage and data networks, they’re also converging siloed IT departments with little overlap in expertise — subgroups that often see each other as rivals competing for the same limited pool of resources. In most organizations, three-headed IT employees with expertise in networks, data storage, and servers simply don’t exist, so companies have to grow their own.
Overall, says Heil, the unified network has been a boon to his staff, which was able to broaden its expertise over time and become a cadre of generalists rather than specialists.
“Originally I thought my network guys would develop a better understanding of the server world, but really the opposite happened,” he says. “My compute people are really much broader than they were before because of the convergence.”
Other organizations might not be so lucky, warns Matthew Brisse, a research director at Gartner. “IT organizations who embrace silos of technologies will struggle with internal cloud and server-hosted virtual desktop projects,” he says. “Technologists need to step out of their comfort zones and work with each other. This means security, server, storage, network, desktop and application groups need to understand the big picture.”
Getting ahead with fewer heads
This is hardly the first time tectonic shifts in technology have upset IT department equilibrium. A similar change happened a decade ago when data and telephony networks converged, says Ajay Chandramouly, cloud computing and data center industry engagement manager at Intel, which is implementing its own FCoE (Fibre Channel over Ethernet) network for storage and data.
“These things tend to come in three phases,” he says. “First the technology arrives. Then you have a debate between the separate camps. As the general level of knowledge increases and both sides become smarter, you can create a generalist role, an overall systems specialist. It’s not just the convergence of Fibre Channel and Ethernet but the larger impact of cloud adoption that is spurring this kind of change. The previously siloed roles of storage, networking, and compute are merging into one cloud systems analyst role.”
Getting there, though, can lead to some tough adjustments for IT personnel. When Walz Group adopted Cisco UCS three years ago, it was able to cut its rack space needs in half, slash failure rates, and virtually eliminate downtime. As a result, Walz was able to able to cut its IT engineering team by two-thirds, says Bart Falzarano, CISO at Walz, which manages critical communications for financial, health care, and government agencies.
The remaining staff attended week-long training courses run by Cisco and Firefly Communications to get up to speed on Nexus switches and fabric interconnects, he adds.
“The current team at Walz is much smaller because fewer IT engineers/administrators are required to effectively manage the company’s technology footprint,” says Falzarano. “To us, converged infrastructure means converging our talents to become more effective with managing, operating, supporting, securing and scaling out our data center technology.”
At Intel, adopting a converged network didn’t lead to headcount reductions, says Chandramouly. Instead, it freed some engineers to work on more valuable projects, while expanding the skill sets of others.
“It allowed us to reassign IT personnel to higher value-added areas,” he says. “We are seeing fewer specialists and more generalists who know about both areas. Improving everyone’s knowledge on both sides is very important. And once you’ve increased the knowledge organizationally, you need to realign the organization so that instead of each team reporting to its own separate operations manager they’re both reporting to one.”
Converging hearts and minds
Adopting a converged network infrastructure also requires a shift in IT culture and mind-set, notes Bob Monahan, director of management information systems at Dynamics Research Corp (DRC), a tech services and solutions firm for government agencies. Among other projects, DRC is working to break down IT silos at the Department of Homeland Security. “There has to be a cultural change where you move away from ‘owning’ the system to owning the data,” he says. “IT personnel have to be comfortable with other entities being involved with their systems and be willing to give up total control.”
It is a natural progression, says Monahan. IT pros have been through it before with virtualization. “The same thing happened with the shift to virtualization — some techs couldn’t handle it because they no longer had 100 percent control,” he says. “Now we’ve gone well past that where we’re dealing with public-private hybrid clouds with elasticity. As that evolved, IT staff evolved as well.”
Moving from siloed storage and networks to a unified system is a much bigger change than shifting from physical servers to virtualized ones, says Paul Lewis, vice president of technology, architecture and security at Davis + Henderson (D+H), a financial services firm based in Toronto. IT managers will need to step up their game as well.
“You need to think about the impact on your existing staff, and how they perceive change, and evolve their thought process as you evolve the technology,” says Lewis. “People who’ve grown up in a single discipline might immediately fear for their jobs. They’ll be afraid you won’t need a storage engineer any more. What you need to bring to the table is that while you might need fewer storage engineers in the future, you will be needing a really good convergence engineer. You need to have a change management mind-set. You can’t have people on your team immediately assuming they’ll lose their jobs, or it simply won’t work.”
Even then, you should plan for a longer period of transition with a few bumps along the way, notes Heil. “Early on in our transition to UCS I was walking out of the office with the person who was in charge of our VMware and Citrix servers,” says Heil. “He was talking about how complex it all sounded. I was thinking the opposite — it all made sense to me. But for the server people the exposure to the storage and networking side was overwhelming initially. Our server people all have a decent understanding of storage and zoning now, but it took a while for them to find their comfort level.”
Boon or bust?
For some, the benefits of convergence can be substantial. By using FCoE software drivers and converged network adapters, Intel doubled network performance and cut the cost per server rack by 50 percent, says Chandramouly.
Gartner’s Brisse is less bullish on convergence, warning if organizations are not careful with their architectural designs, it could end up increasing network complexity without boosting performance. “Data center bridging holds out the ultimate promise of networking — a single network to transport all types of traffic, including storage, Internet Protocol and high-performance compute clustering,” he says. “Unfortunately, interoperable products may never appear, because nearly every vendor has added proprietary extensions. Even within a single-vendor network the benefits are at best limited.”
Despite the challenges, D+H is plunging ahead with convergence. Instead of converting its existing infrastructure, which, thanks to multiple acquisitions, includes a wide range of disparate platforms and technologies, the financial services firm is launching a greenfield network for a new business division, which it hopes to have online by the fourth quarter of this year.
If all goes well, D+H hopes to gradually transition the rest of its infrastructure to a converged environment over time, says Lewis. His advice? Smart small. “Start with an appliance instead of converged infrastructure,” he says. “Doing it on a much smaller scale give you a better appreciation for the people and process change that’s about to occur, some of which will be tough.”
This article, "Manage infrastructure convergence without losing your grip," was originally published at InfoWorld.com. For the latest business technology news, follow InfoWorld.com on Twitter.