(Editor’s Note: This is a multipart series on how companies are rising to meet the challenges posed by the Next Generation Network via advanced management products and services. Unlike the circuit-switched PSTN of old, the new packet-based NGN consists of a multitude of converged services, multimedia and other forms of digital traffic. As traffic monitoring and control become more automatic, attention shifts from the network infrastructure exclusively to maintaining a high quality of service for customers and their Service Level Agreements via performance assurance, traffic prioritization, bandwidth shaping, and related technologies.)
NetQoS offers the NetQoS Performance Center, a web-based management dashboard that gives operators an integrated view of key metrics for network performance monitoring and management such as: end-to-end performance, traffic analysis, VoIP quality, and device performance, all from a single console.
Patrick Ancipink, Director of Product Marketing at NetQoS, told TMCnet in an interview that the company has been around for 10 years, focused on network performance.
“Our tagline is ‘Performance First’ – so we definitely look at everything through a performance lens,” Ancipink said. “Traditionally, the way we’ve served the market is through direct sales to enterprise customers. We have about 1,000 customers right now, and that come about engaging with staffed enterprises that are themselves responsible for assuring applications delivery over their networks. We pin some of our technologies and capabilities to the ‘crossing the chasm’ model, and that’s fairly instructive of where we think we are with the enterprise market and how that’s transitioned into what we’ve doing with service providers. If you start with the concept of managing the network for performance, you need some idea of what normal network behavior is.
“A great way to do that is to baseline what your application performance is, not in an overall, summarized way, but taking into consideration that Monday a 9 a.m. when everyone comes back from the weekend, what does the performance look like then, as opposed to Thursday at 10:30 a.m. So it’s really an important concept to have a definition of ‘normal’ and to have those baselines, you can identify exceptions. The way that we do that is through a response time product that measures end-to-end response times if not synthetic transactions.
“It’s not a sampling of data it’s looking at all of the applications activity that’s going on and building those baselines. That’s how we start with our performance-first approach: define ‘normal’ and the recommended way to do that is via the response time side of the application. You can do that by basically reading TCP packet headers. That in itself is simple, but then there’s a lot of analytics in there to actually come out with what those baselines and dynamic thresholds might be.”
“In addition, we also look at the actual network traffic over the links,” says Ancipink. “Cisco’s (
News -
Alert) NetFlow is the primary source of data for that. Our company’s growth really took off around 2004 when we were persuasive to get a lot of people to turn on NetFlow and use it as one of the sources to see how the network was supporting applications delivery and to understand which protocols are the most popular and who are the top ‘talkers’. To that you can add SNMP and device statistics. Many have relied on those solely and they think they’ve been doing performance management.
“We think it’s an important ingredient but it has to be included in the context of those other data sources I’ve mentioned: response time, traffic analysis, and so forth. There’s a recurring theme here – no single data source is really adequate to understand overall network performance. And then you start adding in things such as voice and video, where definitely up/down status is meaningless. It’s all about, ‘What is the quality of that interaction or conversation?’, where jitter and latency basically don’t represent just degraded performance but more or less no performance – even though something may be available, if you can’t hear the person at the other end of the phone, what’s the point?”
Ancipink adds, “Although we have a specific product that gathers call quality statistics and other voice metrics to calculate MOS [Mean Opinion Score] data, it’s also important to have information such as, what are the other protocols on that converged network that are competing with voice, and how to do you make sure that can prioritize traffic properly according to what your needs are. So all of these things used in combination is really how we’ve worked with our customers so they can have an overall view of what’s going on.”
“We monitor QoS but we don’t actually do the control side of it,” says Ancipink. “We do provide you with all of the information that gives you the decision support of how to configure QoS and then assure or validate that it’s working. Whether that’s ensuring a specific application or trying to figure out which sites are good candidates for WAN optimization, we’re determining whether something is a server problem, an applications problem or a network problem. We can bring back all of that information so that you can make proper infrastructure investments or QoS configuration – anything along that line.”
Mike Magri, Client Delivery Executive for Service Providers at NetQoS, adds, “One of the things our customers really like about our approach across all these data sources is, when you’re doing something in IT, whatever that workflow may be, whether it’s capacity planning, troubleshooting, or you’re looking at migrating a data center or server virtualization, or whatever it is, you need different data. Any one of those data sources doesn’t necessary provide you with everything that you need, so you can go across these data sources and build your workflow in our portal such that it doesn’t matter where the data comes from or what the capability was that collected it, to give you exactly what you need to do capacity planning and to determine what the usage is, or how well you’re delivering voice services, so you that you can go across that via workflow.
“So there are multiple data sources or different ways to collect it, but it all has to be pulled together in some meaningful way. There’s a lot of ‘swivel chair’ engineering going on with different toolsets, often with disparate or competing data. We’re trying to reduce that as much as possible by helping people to standardize on a toolset, one that can be shared with a server or an application team as well. It speeds troubleshooting and reduces a lot of the finger-pointing as well.”
Ancipink then elaborates about market adoption, “We have about 1,000 customers, over half of which own multiple products from us. That metric keeps growing and it validates our strategy with the way we build the portfolio. When we look again, referencing back to the ‘crossing the chasm model’, we feel like we’re hitting perhaps the beginning of mainstream market. We’re seeing the types of companies or organizations that are laggards in technology adoption now opening up an embracing having a way to measure all of those shades of gray regarding network performance, and getting off of the utilization or availability crutch, in the process adopting the change in culture and processes that are required.
“There are several interesting things about this. One is that many of those ‘laggards’ don’t acquire technology the same way the early market folks do. That’s where service providers become important. From the service provider side, they’re being asked to provide more differentiation and more visibility. So we see opportunities on both sides. We’re approaching a market that’s not traditionally serviced by direct vendor sales. They don’t necessarily have the team that can acquire and operate that level of technology, or they choose not to or perhaps they are a relatively small organization.
“Moreover, the people who supply them are very receptive to adding differentiation and more offerings – more ‘things’ that they can sell. That’s where we are right now – we feel that we’ve hit that inflexion point of where enterprises ‘get it’ and we’ve sold product to a large number of them and we’ll continue to do that; but the next phase of hitting the market has been harder to achieve with this multiple data source, high visibility approach that changes their position. But that part of the market is growing for us.”
Mike Magri continues, “My job at NetQoS is to be the worldwide overlay for our service provider and integrator customers. As Patrick Ancipink was talking about in terms of market adoption, we’re seeing a good deal of those customers who are ‘laggards’ on the large enterprise side, and small and medium enterprises that are looking to get into performance evaluation and assurance, are all turning to their existing vendor relationships, saying, ‘Hey, what can you give me?’
“Additionally, in the migration of MPLS [MultiProtocol Label Switching] across all segments of the industry, they need their visibility back, because what happens is that, when you adopt MPLS, you’re turning all of the control over to your carrier. Customers are now going to their providers, saying, ‘I need more visibility into what the traffic is and how well it’s being delivered. So what we’re seeing across our customer base – our service providers – is that they’re looking to offer their more mature customers who want these services visibility through NetFlow, and active testing of loss and latency on their networks via IP SLA which allow administrators the ability to Analyze IP Service Levels for IP applications and services.
“Providers are looking to offer their customers application performance monitoring of such things as SAP and Peoplesoft and email. And then, as we approach voice and video, now we’re talking more about emerging technologies that are not quite as well-formed along the Geoffrey Moore’s ‘crossing the chasm’ model, and so for certain companies that are providers of voice and video to their customers, their ‘leading edge’ is such that not only do they neet to show performance at the infrastructure level, but they also need to show quality of experience for voice and potentially for video and unified communications.”
Providing a Superlative Experience
“So we’re seeing that some providers are just saying ‘I need visibility’ or ‘I need application performance”, but other providers are saying, ‘I need to go much more to the leading edge and give quality of experience for VoIP or unified communications’,” says Magri.
“In terms of what we’ve got going on right now, I can think of two specific examples of these differences that I was talking about,” says Magri. “The first of these models concerns a St. Louis-based company named Savvis, Inc., a leader in outsourced managed computing and network infrastructure for IT applications. Savvis has built a visibility service that can show you all of your traffic, link-for-link on your MPLS network. And then, if you have packet loss and latency-sensitive applications such as voice and video, they also will do IP SLA testing between sites so that you can know what your baseline is and what’s ‘normal’ performance, and you can get ‘alarmed’ and will receive notification any time performance gets outside the bounds of acceptable service delivery. So Savvis is just now launching their service at this point in 2009. We’ve already helped build it and we’ve got a few pilot customers. It’s ready to go to market.”
“The second example is much more out on the forefront of voice, video and quality of experience,” says Magri. “The NetQoS customer I can mention here is eLoyalty. They help with call center performance, and their Call Center Managed Services group, CCMS, uses our NetQoS Unified Communications (
News -
Alert) Monitor. They use it to monitor call performance for their customers and charging each customer to provide them with those type of reports and that kind of monitoring. So in this case it’s not visibility service, but a quality of experience reporting service.”
To be specific, NetQoS Unified Communications Monitor is a network-based call setup and call quality monitoring product that tracks the call quality users experience, provides alerts on call performance problems, and isolates performance issues to speed troubleshooting and mean time to repair. The product enables network engineers and IP telephony management to assess the performance of their Cisco Unified Communications Manager (CallManager) IP-PBX (
News -
Alert) by tracking, evaluating, and reporting on key metrics without deploying server agents or probes. For every call, Unified Communications Monitor reports on user call quality and the underlying network performance metrics associated with it. Unified Communications Monitor also breaks out performance data from the IP and PSTN legs of calls that pass through voice gateways traveling to endpoints in the PSTN. This data lets IT organizations know if the network is responsible for less-than-optimal call quality.
Please find part one of this series
here, part two
here and part three
here.
Don’t forget to check out TMCnet’s White Paper Library, which provides a selection of in-depth information on relevant topics affecting the IP Communications industry. The library offers white papers, case studies and other documents which are free to registered users.
Richard Grigonis is Executive Editor of TMC (News - Alert)�s IP Communications Group. To read more of Richard’s articles, please visit his columnist page.Edited by
Michael Dinan
