When Cardinal Health Inc. replaced its legacy network monitoring tools last year, the multibillion-dollar health care services company hit the network analysis sweet spot—the new software offered the ability to monitor both network hardware operations and application performance, while providing analytics that could improve overall business processes.
Cardinal’s investment reflects a shift in the IT industry where information technology directly affects business intelligence and enterprise productivity -- and the network plays an increasingly central role. The idea is that the network ties together every element of cloud provisioning and service or application delivery. To handle this, monitoring tools must provide a broader set of information.
“Networking used to be pretty much its own little silo,” said Jim Frey, vice president of research, network management, at Enterprise Management Associates. “Now we’re seeing a shift where networking is bringing with it systems, applications, storage and other components, providing a more systemic view. Organizations need to see a cross-domain view of operations. If something is broken in one domain, they need to see the problem in another domain.”
Now that the network has become “relevant again,” new monitoring strategies are emerging, said Dimitri Vlachos, Riverbed Technology’s vice president of marketing and products.
“There are finally tools for network teams to see the role the network really plays, and that it’s an integral part of [ensuring] app performance across to end users,” said Vlachos. “It’s not just, ‘Hey, I’m going to look at my network for the network’s sake.’ It’s really diving into how well it’s delivering applications.”
Cardinal found its network monitoring answer in SolarWinds’ Server and Application Monitor and Network Performance Monitor applications, placing eight polling stations running on Windows-based virtual servers between two data centers for redundancy. The software monitors 10,000 devices in two data centers and 148 satellite offices.
“We brought [SolarWinds in] to handle availability monitoring, hardware monitoring, what I call, shrink-wrap app monitoring and custom app monitoring,” said Leon Adato, Cardinal’s monitoring architect. “‘We have a specific app? Let’s monitor that.’ This type of granular examination wasn’t possible with the older system.”
Tools that align with business process
Cardinal can use this clearer picture of application performance to improve business processes.
“We can go back to the development teams and ask them what they really want from their application performance in the first place,” said Adato. “We will be able to go from ‘alert convert’ to ‘alert improve.’”
Specifically Adato’s team can more accurately monitor the performance of custom applications that drive Cardinal’s business.
“This will give us the ability to drive up the chain from hardware to the OS to the app itself, to get us to what I call business-process monitoring. I want to be able to say things like, ‘I see 35 connections to the Web server; 15 are on the sales page but only one is going through. Why?’ We never had that type of information before,” Adato said.
For the IT team, the positive outcome of the new tools is already clear—performance data is more precise, he said. “Our Exchange team went from 80 [help desk] tickets per month to eight, and we know now those eight tickets are real; the other 72 were noise. When [the support team members] get a help desk ticket at 2 a.m., they know they have to get up and move.”
Next up for Cardinal: virtualization and voice-quality monitoring. Adato is currently reviewing vendors that include SolarWinds, VMware and others.
A patchwork of monitoring tools
At Oppenheimer & Co., the New York City-based financial services organization, Henry Jiang, executive director of network systems and IT infrastructure, engineered a network monitoring strategy that links together a number of tools for packet inspection, server monitoring, network uptime and anomaly detection.
With only seven employees in the company’s IT infrastructure/network services team, Jiang’s greatest “challenge is increasing operational efficiency,” he said.
The department oversees two data centers and 100 branch offices, spanning more than 2,000 devices. Responsibilities include the data centers, the network, IP management, engineering support and data security. “We are heavily reliant on different toolsets that let us pinpoint problems from an operational perspective,” Jiang said.
More on monitoring networks and applications
Three vendor strategies for application performance monitoring
Monitoring in VXLAN environments
How to choose server monitoring tools
Jiang’s primary monitoring tool is Riverbed’s Cascade network performance platform, but he also uses products from Ipswitch Inc., Net Optics Inc. and SolarWinds for uptime, packet inspection taps and server-monitoring, respectively.
“From the networking side, I’m getting huge benefits [from Cascade],” Jiang said. “It’s capturing data flow from the entire network, and that’s extremely important.” Jiang uses Cascade’s software to create long-term versus short-term performance charts that allow network operations center (NOC) workers to see network traffic patterns at a glance, which reduces the amount of time operators must spend assessing traffic. “They just take a look and they can determine if we are operating in a normal parameter or if there are anomalies,” he said.
Oppenheimer employs a hub-and-spoke network topology, which sends traffic from branch offices to the New York-based data centers via Multiprotocol Label Switching and dedicated links. “This model lets us detect any traffic anomalies [including viruses and other threats] because everything comes back through the chokehold of the data center,” Jiang said.
Within the data centers, Jiang’s team gathers NetFlow data generated by the centers’ Cisco equipment, but performance information is also captured by the Net Optics physical taps. “This fills in the Transmission Control Protocol performance parameters, server delays, network delays. These are the parameters that can only be detected by looking at the actual packets.”
The monitoring foundation gives Jiang the performance and status information to ensure uptime and data integrity, he said, “IT here is a function; we provide a service to the business. And we have to do more with less. For example, the NOC. It’s a key function that I manage as a primary responsibility. That NOC is a single person operating on his or her own all day. Without tools, I would need two shifts, two people, so the toolset I use is extremely important to streamline that process.”
Vendors fortify monitoring platforms
Network monitoring vendors are responding to the more complex needs of their users with platforms that span hardware and applications, integrating information from both. NetScout Systems Inc. and Compuware Corp., for example, each recently rolled out software that allows customers to gauge both application performance and network operation.
NetScout’s nGeniusONE, unveiled in late June, tracks the delivery of services and their underpinned applications across a company’s entire infrastructure, said Steven Shalita, vice president of marketing. The deep packet inspection app supports data, voice and video, as well as remote desktop applications and unified communications services. Users access performance data through any HTML5 browser.
NGeniusONE examines all other components within a service, from software and firewalls to routers and protocols, to determine performance and ultimately user experience, Shalita said. “Any one of those components could be giving you a problem, but if you are just monitoring the network or just the app or the application delivery controller, the light could be green but the downstream performance could be suffering.”
All this monitoring and reporting occurs in real time. “We’re not just collecting traffic and analyzing later. We can from a dashboard perspective tell you how a service is performing and give you an early warning if there is a problem,” said Shalita.
Platforms like these are all part of the broader focus on application behavior, said Patrick Hubbard, head geek and senior technical product manager at SolarWinds. “Five years ago, you knew which apps were being added [to a network] and you could build something that would allow the traffic to flow and fulfill the business requirement,” he said. “Today, the network team is providing software from a Web server. Administrators are looking at traffic like never before; they need to know what’s filling up the pipes. It’s more than QoS [Quality of Service] or ToS [Type of Service]; it’s using quality maps to do more thoughtful and intricate packet shaping [and thus optimize application delivery] than in in the past.”
Networks, jobs become more complex
Creating network monitoring tools to meet these demands is becoming even harder now that network architectures are getting more complex as a result of virtualization, software-defined networking (SDN) and the converged data center.
“I’m virtualizing servers; I’m overlaying networks; I have an SDN-defined data center. Within the data center, I can have all of these virtual data centers running on top of it. How can an operations team keep track of all these systems for each customer? How can I understand the performance they are experiencing and what problems they are facing when they occur?” Riverbed’s Vlachos said. “It used to be, ‘It’s not me.’ Now it’s shifting to ‘We are part of a team that will solve the problem.’”
At this transition occurs, network engineers may have to expand their own skill sets. “The notion that you will have dedicated network people focused only on networking, those days will be gone,” Vlachos said. “As we move forward, you will have to understand more about apps. You will have to script, and you will be automated. It’s all about an interconnected world.”
About the author:
Chuck Moozakis is the site editor for SearchNetworking.com. Chuck has covered networking, telecommunications, new media and newspaper and magazine production technologies for more than 25 years. Prior to joining TechTarget, Chuck was editor-in-chief at News & Tech and also served as senior editor for InternetWeek.