Wide area network (WAN) monitoring tools typically blast WAN managers with alarms about degraded WAN application...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
performance and leave it to the engineers to do the troubleshooting. Although many WAN managers prefer having manual control over troubleshooting, some IT organizations are using appliances that automate the process in order to make WAN application performance management less labor intensive.
"It's a waste of resources to have someone sit there to just watch what's going on," said Andre Kindness, senior analyst at Forrester Research. "People are [demanding] more automation for traffic and control."
Thomas Masurel, CTO of Moët et Chandon (M&C), the wine and spirits division of French luxury goods conglomerate Louis Vuitton-Moët Hennessy (LVMH), has outsourced day-to-day WAN management to AT&T, leaving just two networking pros to oversee a WAN spanning 150 production and distribution sites and 6,500 users worldwide.
Remote users had increasingly complained about WAN application performance as M&C centralized more and more of its applications into its Montpellier data center, Masurel said. Web surfing and email traffic, which accounted for about 60% of bandwidth usage, were stalling the performance of business-critical applications. The performance complaints had Masurel's networking team consumed with WAN application performance management.
“[Manual troubleshooting means] you spend a lot of time for only a few percent [higher] amelioration of quality of service," Masurel said. "We don’t have enough people to do management and fine tuning, so I prefer automatic solutions."
In theory, carriers can ensure that critical applications get priority across the WAN by applying quality of service (QoS), but Masurel said the "totally static" nature of that approach is unappealing.
"You can ask your provider, 'OK, on my bandwidth, I just want to keep 5% of it for SAP, 10% for BI, 30% for Web and 20% for email,' but then you have a static solution," he said. "If, for example, there is no Web surfing at one time, then the bandwidth you just reserved for Web surfing will not be used."
Since December 2009, M&C has used Ipanema Technologies' WAN optimization and management platform to automate WAN application performance management and troubleshooting through dynamic QoS. Instead of calculating and assigning fixed bandwidth allocations, Masurel assigns applications to different QoS objectives, or what Ipanema calls levels of "criticality" (top, high, medium and low).
Ipanema software ships with a catalogue of performance profiles for popular business applications, according to Peter Schmidt, CTO of Ipanema's North American operations. The appliance can also learn how custom applications should perform, he said.
Depending on WAN conditions, the criticality level that applications are assigned, and performance thresholds -- measured by server response time, packet loss, jitter, TCP retransmissions and more -- the Ipanema appliance automatically shapes traffic to prioritize the highest-priority applications, Schmidt said.
"The main strength of Ipanema is their business use [perspective] of the network," Masurel said. "They are not optimizing the network performance. They are optimizing the application performance -- using the network."
Blue Coat Systems' PacketShaper appliance offers a similar approach -- preloaded with or learning acceptable application performance and using QoS to maintain those norms accordingly.
Meanwhile, other vendors have embraced an automated and dynamic approach to WAN monitoring but don't automate troubleshooting. NetScout Systems and Riverbed Technology offer appliances that "know" or "learn" average or acceptable performance of a given application and alert administrators when that threshold isn't met.
Automating WAN application performance management: Too good to be true?
Although the concept of automated WAN application performance management appeals to IT pros, some are skeptical about vendors' ability to execute and prefer the passive approach for now.
[Automated WAN application performance management] would certainly be wonderful, but then again, I think in that situation you might end up making compromises.
Sean M. Power, IT Manager for Security and Infrastructure,
Lathrop & Gage LLP
"That would certainly be wonderful, but then again, I think in that situation you might end up making compromises," said Sean M. Power, IT manager for security and infrastructure at Lathrop & Gage LLP, a Kansas City-based law firm with 12 sites and about 600 users. He has never used Ipanema's products.
Power said the automated reporting and alerting capabilities in Riverbed's Cascade WAN monitoring product -- which he uses alongside its flagship Steelhead WAN optimization appliances -- have benefited the IT team greatly and not overwhelmed them with alarm bells.
"Our concern was that we realistically didn't have a chance to sit and define absolute thresholds for various applications," he said. "It has automated analytics that actually learn the real behavior of the network and then alert you when there's a deviation from the norm."
Having recently invested in a security information and event management (SIEM) platform with remediation capabilities, Power hopes to eventually integrate it with Cascade so that the two can automate WAN application performance management and troubleshooting.
But the project is a long way from testing and production, he acknowledged. He would rather see Cascade further integrate with Active Directory to support Windows Server 2008 R2 so that identifying WAN application performance issues and user behavior becomes easier.
Meanwhile, WAN monitoring vendors fire back that the notion of a self-healing WAN -- at least in terms of application performance -- is probably too good to be true.
Automating management of WAN application performance is complex and difficult to achieve, said Steve Shalita, vice president of marketing for NetScout, which sells nGenius, a passive network and application performance management product.
"There are a lot of things you can automate,” Shalita said, “but understanding the relationships between the applications and getting into the user experience is extremely complicated and extremely difficult to do in large environments.“
Yoav Eilat, product marketing direct of Cascade at Riverbed, said that automating application performance management needs to look beyond the WAN and into the data center for potential hardware problems. He said Ipanema has limited visibility into data centers.
Schmidt said that Ipanema has "carefully designed the system … from the beginning with full visibility, unlike other vendors who came to the visibility party late and are modifying a security product."
Let us know what you think about the story; email: Jessica Scarpati, News Writer.