Wide area network (WAN) engineers who focus on infrastructure are missing the big picture, analysts warn. Everything...
should be framed in terms of applications, not routers. WAN professionals need to know how to develop and deliver on internal service-level agreements (SLAs) for critical applications. Everything else is just gravy.
The good news in this paradigm shift is that not all applications are created equal, even if end users think otherwise.
Identify that handful of applications and ensure that they are in top shape while delivering reasonable service on the rest, Metzler advised, and IT departments will see happier users and will align more closely to business objectives. That latter factor becomes critical when budget season arrives. The WAN team will be able to demonstrate the value it creates and not just the dollars it consumes.
Identifying critical WAN applications
Identifying critical WAN applications is key, and these can come in all shapes and sizes: point-of-sale terminals that use only tiny amounts of data, electronic trading platforms that are not only high-bandwidth but also super latency-sensitive, and software as a service all might demand continuous uptime for a business to succeed, but few businesses have exactly the same needs.
Unfortunately, asking various user groups what is critical may or may not be the best way of figuring out which applications are most important.
In a recent study Metzler conducted on behalf of NetQoS, a network performance monitoring vendor, 41% of respondents said that they queried business managers about critical applications.
"You're damned if you do and damned if you don't," Metzler said. All too often, he warned, when asked about "critical" applications, business managers tag every application they have.
The upside is that getting business manager input can show IT alignment with corporate objectives, so he recommends seeking out the input but taking it with a large grain of salt.
Metzler suggested that a more accurate monitor is listening to user complaints.
"See what you get beaten up for not working," he said. Help-desk tickets can be one indicator, but so can the verbal thrashings that come in over the course of time as applications invariably fail at one point or another.
Another method is to look at network traffic, at least to identify the applications that are out there.
Such a view can often result in surprising insights because business managers sometimes do not even know that a critical app they rely on exists until it goes down.
"It seems every time we talk to a customer and give them full visibility, they're really surprised about what's out there," said Patrick Ancipink, director of product marketing for NetQoS.
Ancipink recounted how the Navy, for example, once contracted an outside vendor to manage 1,000 applications on its network, only to discover that it was running 10,000 applications.
Regardless of the total number, however, Metzler said it is critical to narrow that range down to the absolutely necessary ones and focus on them.
"Culturally, it's difficult for IT organizations to do that," he admitted. "They say, 'Jim, in our organization, everything's the highest priority,' but that also means everything's the lowest priority."
Create a secret WAN app SLA
Ultimately, the goal of identifying critical applications should be to create and maintain an SLA with business units to show, in indisputable terms, where the IT investment is going. Jumping right into that can be a risky proposition, however.
"Over-promising and then coming up with an SLA you can't meet because it's too complex to do isn't good for anyone's career or the reputation of the IT organization," Metzler warned.
Instead, a secret SLA might be the way to go, at least as WAN managers test what their groups are capable of delivering reliably.
"I often advise IT shops to identify one or two business-critical applications, develop an SLA with it, and don't tell anyone," he said.
Moving past IT silos
Ultimately, the move to a more SLA-centric WAN future is a step toward breaking down IT silos, Metzler said.
The networking teams, applications teams, disaster recovery (DR) teams and more will all work in concert, he predicts.
"The network becomes a means to an end and not the end in itself," Metzler said. "Now, my goal might be the uptime of the WAN. As long as I've met the uptime of the WAN, great. In the new world, I have the uptime of the WAN plus the performance of 50 or 60 applications."