abednarz
Executive Editor

Organize to automate

Feature
Oct 25, 20049 mins

New data center technologies are offering IT shops a surprising benefit – improved cooperation.

Automation know-how
Outsourcing meets automation
Organize to automate
Grid on the job
Danger in the new data center

Open floor plans – where kitchen, dining and living rooms meld together – are in vogue today among home buyers who’d rather have one large space than three distinct walled rooms. Analysts suggest a similar trend in IT departments.

Traditionally, companies relegate IT staff to stovepipes, divided by specialties such as applications, databases, storage, servers and networks. But emerging technologies are spurring companies to question whether greater interaction among these distinct disciplines might be warranted.

One reason is that many of the new technologies making their way into data centers are more cross-silo than previous tools. Current monitoring tools have more of a business-like orientation than their specialty predecessors, for example. They’re less concerned with the performance of a single component and more concerned with finding the root cause of a problem and reporting its effect on business processes and end users. “Today’s software understands relationships and interdependencies between network, server and application components,” says Trent Waterhouse, product marketing vice president at management software maker Aprisma.

Taking advantage of the feedback these systems can provide requires IT departments to rethink how they traditionally purchase and use technology. “People and processes have to be cross-silo before the tool can be effective,” Waterhouse says.

Similarly, vendors with virtualization technologies propose to help companies reduce the number of physical servers in their data centers by combining multiple instances of servers and applications on each box. However, really digging into the benefits of such technologies sometimes requires cross-discipline efforts.

Server virtualization technology is changing the way some staffers interact at Gannett, says Eric Kuzmack, IT architect at the century-old media conglomerate in Silver Spring, Md. “I don’t think virtualization is changing the relationship between application and infrastructure people. To the application guys, it’s still just a server. They have no idea that an application is running on a virtual machine,” Kuzmack says. However it could spur the relationship between the server and network teams to change over time, he says. One reason is a feature like trunking. A virtualized infrastructure lets servers take advantage of network trunking, where one server can be linked to multiple networks via a single cable.

But routers usually support trunking, not servers, Kuzmack says. Taking advantage of the feature requires collaboration between Gannett’s network specialists – who are familiar with trunking requirements – and server folks.

Some technologies naturally seem to strengthen the alignment of IT disciplines. For example, application traffic-optimization products blur the lines of networking and applications with features from both disciplines working in tandem.

The migration to Web-based applications has made application specialists more concerned with optimization and security issues, says Raj Kanaya, vice president of product strategy and alliances at NetScaler, which sells devices that can operate as Layer 4-7 switches, security appliances and Web traffic accelerators. These days it’s common to see network and application staffs at the buyer’s table. “Lots of customers today are thinking cross-functionally,” he says.

Getting the most from the NetScaler appliance requires collaboration, agrees Mano Sivashanmugam, vice president of technology at uBid.com, an online auction site that uses load balancing, compression, Secure Sockets Layer acceleration and site security functions. The network team works with database and application specialists to understand what they require, then builds the appropriate security and traffic optimization rules into the device.

Territorial constraints

Unfortunately, new technologies don’t always improve communication among IT groups. In some instances, technology can almost impede communication. For example, a management tool that hones in on a performance issue and identifies an overloaded application server or a missing database index eliminates the need for finger-pointing among IT groups, which is a good thing. But it also lessens the need for interaction, says Steve White, manager of systems engineering at PacifiCorp. in Portland, Ore.

The electric utility uses Veritas Software’s server clustering technology to move applications and services automatically from one available server to another if a failure occurs or service degrades. “If there’s one tool that everybody could go to and look at the red light to find the problem, that’s going to give you speed and efficiency. But it doesn’t require more information sharing among groups, since the tool does that for you,” he says.

Information sharing is necessary and becoming more important early in projects, White says. Application developers are recognizing that the design of their applications – thin client vs. thick, for example – should take into account how the infrastructure is laid out, he says.

“Information sharing has great value in the architecture and design processes,” White says. PacifiCorp has well-defined, multi-disciplined project-delivery processes, but there’s room for improvement. “Lately there’s more up-front cooperation and collaboration, especially around design issues,” he says.

While experts agree collaboration is on the rise, many are skeptical that job descriptions will change dramatically, even as new data center technologies take root. Users say the distinct skills required of an application expert vs. a network guru aren’t going away. “I don’t think we’ll see those skill sets melding, especially in larger IT shops,” White says.

Seven cooperation tips

Experts share advice on how to best begin breaking down IT divisions.

1. Stress collaboration. Application, security and network teams must get together in a project’s early stages to ensure “everyone understands the performance requirements and each part of the design fits together to meet those requirements,” says Steve White, manager of system engineering at PacifiCorp in Portland, Ore.

2. Choose the right staff. Begin cross-silo initiatives with the network staff, says Trent Waterhouse, vice president of product marketing at Aprisma. “The network ops center has more experienced people, better documented processes. The instrumentation is already standards-based” and the network touches every system, he says.

3. Look for broad-thinking IT managers. The need for technology specialists isn’t going to go away. The key is finding managers who see the big picture and can manage across silos, says Rakesh Patel, a vice president at Windward Consulting Group.

4. Adopt common tools. Get different IT teams to use some of the same tools — for example, systems monitoring tools, Patel says. But don’t underestimate how difficult it is to try to integrate all these tools, he warns.

5. Don’t go overboard. Redoing the entire IT organizational structure is unrealistic, says Corey Ferengul, a senior vice president at Meta Group. “When you start talking about a dramatic change that’s cultural in nature, no one actually pulls that off,” Ferengul says.

6. Start with process. Consider designing processes that link IT groups, Ferengul says. “This doesn’t eliminate the groups, but it pokes some holes between them,” he says. “Ultimately you’d love to see IT teams sharing data and technologies, but if they at least have shared processes and know who’s supposed to do what, and when the handoffs are supposed to occur, you’re in a lot better shape.”

7. Think career enhancement. A career-savvy network manager could look at a silo-busting initiative as an opportunity to work with a CIO to tackle business-service issues, Waterhouse says. This is the path for career advancement, he says.

— Ann Bednarz

In some ways, virtualization technologies might even heighten some of the differences between skill sets, says Michael Mullany, a vice president at VMware, an EMC business unit. “A virtualized infrastructure separates management of software from hardware,” he says. So to the applications team, the hardware platform is almost irrelevant. “A virtual hardware platform makes everything look the same as far as software is concerned.”

The stovepipes will continue for a while, says Rakesh Patel, a vice president at Windward Consulting Group. It would be difficult to find staff members who have skills across a wide breadth of tools, Patel says. “It’s hard enough to get a Unix guy to learn Windows, never mind routers and switches,” he says.

Another big obstacle is pressure from users to maintain stovepipes, says Steve Tuecke, chief architect of the Globus Toolkit at the Globus Alliance. A pioneer in grid research and development, Tuecke advises enterprise user groups looking for ways to spur adoption of grid computing in corporate environments.

A common challenge is getting user groups past their worry that sharing resources would disrupt the performance of their systems, he says. Equity traders might reject the idea of assigning computing power sitting idle during their off hours to other projects because they fear problems occurring at night would affect them in the morning, for example.

These social hurdles are very real, Tuecke says. “Until users believe the infrastructure will be there for them when they need it, they’re not going to give up control,” he says.

Looking ahead

Setting reasonable expectations is important, analysts say.

Everyone talks about breaking down the silos of IT disciplines, but that’s not entirely possible, says Corey Ferengul, a senior vice president at Meta Group. “The reality is you can’t break the silos, because you still need the low-level technical knowledge that comes out of each group. What you can do is poke holes in the silos so that the different IT teams see each other better and work better together,” he says.

Keeping everyone on the same page is important. Meta Group uses the term “aggregation layers” to describe people whose role it is to aggregate input from business leaders and IT teams “and make sure that those groups see eye-to-eye without necessarily having to talk every day,” Ferengul says.

UBid.com hasn’t needed to add an extra layer of management oversight to make sure IT stays on task. But Sivashanmugam and his team do make sure IT objectives stay in-line with business priorities.

“IT is a manifestation of business strategy,” he says. “I’d like to see all the groups working together to support the business strategy – which is to keep costs low.”

abednarz

Ann Bednarz is the executive editor of Network World. Ann is a longtime IT journalist and has spent 26 years writing and editing for Network World, where she has worked as a news reporter, managed product testing and reviews, and developed features and how-to articles for an audience of network professionals and data center managers. Over the last two years, she has conceived and edited award-winning content for Network World that includes 2025 Jesse H. Neal Award finalists, 2025 Azbee Award regional winners and national finalists, and 2024 Eddie & Ozzie Award finalists.

Ann holds a bachelor’s degree in architecture and spent the early part of her journalism career writing about architectural design and construction. In her free time, she keeps those skills alive through DIY projects.

More from this author