Whether they like it or not, wide area network (WAN) managers need to look beyond the network layer and provide...
WAN application performance management for key business applications. In a stubbornly flat job market, shrugging off application performance issues as someone else's problem is not going to impress CIOs looking to dump dead weight from IT organizations.
Paul Brady, former president and CEO of Mazu Networks, now vice president and general manager of the Cascade business unit at Riverbed Technology, spoke to SearchEnterpriseWAN.com about how WAN management and monitoring tools can help networking pros make sense of application performance issues and identify problems before users flood helpdesk with complaints.
How is the role of WAN management changing?
Brady: I think there's been a strong [transition] from troubleshooting and what lots of our customers call "mean time to innocence" to performance management. Once upon a time, the various IT silos or functions spent their time proving it wasn't their fault, and that's no longer good enough for the CEO, CIO or the business owners. Performance issues are very disruptive [and] they're expensive. If IT organizations can't prevent or at least very quickly fix disruptions in performance, they are costing their businesses money. That's been the biggest change.
I think the other thing compounding that is there's this proliferation of endpoints from smartphones to laptops to iPads and also a dramatic increase in applications that are being used over the network. Once, you could instrument your network understanding what it would look like for the next three, six or 12 months. Well, now, the minute you instrument it, it changes. The [WAN application] performance tools required to succeed in today's environment need to be adaptive and dynamic, and they need to be malleable to change as the network induces the changes.
Why should WAN managers care about WAN application performance?
Brady: I think it's been because a) that's what the business demands, and b) once upon a time, as I said, the network team would simply say, "It's not our problem," and that was good enough for them to move on. Now, I think the business and the CIO don't want to know that it's not their problem -- they want them to solve the problem. I think there [has to be] convergence organizationally or the teams need to [collaborate] -- the network teams and the application teams -- because at the end of the day, it's too expensive to have these silos that all declare it's not their problem. The reference point for these guys is now solving the problems to provide the business services [that] the network and applications are there to [provide] -- not simply to provide an infrastructure and declare it stable.
What are some of the key differences between understanding local area network (LAN) and WAN management?
Brady: Let's say you have three LANs, and there are conversations happening across LAN 1 and LAN 3, and you don't have an end-to-end view -- you have a view of LAN 1, LAN 2 and LAN 3 -- you have to do a lot of analysis to speculate what the transactions, conversations or network traffic looks like across [the WAN]. We provide this network-wide view, so we basically roll up those conversations in that network traffic to give you a top-down view, which I think is much easier for [WAN managers] to understand than to piece it together. "I saw this traffic leave LAN 1, and I think I see this traffic coming into LAN 3. Is that traffic related?" If you instrument it at the LAN level, that’s what you're left with. If you have this top-down, Netflow-based view, you're seeing the conversations across the network … so it looks like a single conversation, and you're not trying to piece them together.
After you've identified the bottlenecks, what's the long-term value of WAN management tools?
Brady: The long-term value would be ongoing performance management because the minute you fix something, something else changes -- usage, applications, network changes. You can look at it two ways: one, a reactive standpoint where there's a problem and you bring in a tool like Cascade to help you solve your problem. Well, more of those problems are going to keep coming and, probably more importantly, there's a proactive aspect of it. One of the features of Cascade is something called analytics, so basically we're able to build a model of what normal behavior is for users, for applications, for dependencies. And once we build that model, we're able to indicate or alert IT teams when there's a significant change. The old way of doing that was to set thresholds, but the problem is when you ask IT guys to set thresholds, saying, "If my server goes beyond 50% utilization," they just don't know what the thresholds are. But with our analytics, you can just say there's been a significant change and you can instantly drill down, and that allows teams to be really proactive and head off problems.
We have a customer that's a major broadcaster, and their vision is: “a) I don't want the phone to ring, b) if the phone rings, I don't want [users] telling me what the problem is. I want to say, 'I've already seen there's been a significant change. It's affecting your performance and here's what we're doing about it.'"
Have enterprises made WAN application performance a priority, or are networking pros still focused on uptime?
Brady: Change often comes out of necessity, and I think the shift from troubleshooting to performance management, which is the language they're using, is an indicator that enterprises need to go about this a different way -- and I think that they are.
Two years ago, we were educating people on the value of Netflow. I think we're no longer doing that. We're really attacking this notion of top-down, proactive, dynamic analytics, and I think you'll see us continue to expand our performance management portfolio with some of the releases that are coming with continued focus on business service management. The way a lot of tools look at it is they give you bits and bytes. The way Cascade looks at it is it's a service-oriented view, so the CEO, CTO or even network operations or engineering teams -- they want to look at it and say, "Is my VoIP working OK? Is my SAP working OK? Are my Exchange servers working OK?" And if it's not, they want to know and quickly drill down to see where the issues are. We'll be doing more with the technology to make it easier and easier, and more business dashboard-focused.
Let us know what you think about the story; email: Jessica Scarpati, News Writer.