Answer the APM call: Apply big data and IoT analytics to WAN optimization

Read about big data methods that can help net admins facilitate end-to-end visibility of network and application performance, which is the underlying call of application performance management (APM).



For the past 15 years, the world of the network administrator has been circumscribed by whatever happens behind the corporate firewall. Likewise, network metrics have been developed around local area network (LAN) throughput, bandwidth consumption, resource utilization, and so on.

Now, the adoption of cloud-based and Internet-dependent applications has changed the game forever. The result is a desperate struggle for most IT network organizations to deliver the end-to-end service levels and performance metrics expected for Internet- and cloud-based applications that operate beyond the firewalls of corporate LANs.

One major knowledge gap is what goes on with network traffic and application performance once an application starts operating beyond the enterprise network. If there is latency, dropped packets, or any other type of disruption that impedes app performance, it naturally becomes a network concern.

Enter the world of big data and the Internet of Things (IoT). I recently spoke with Ajit Gupta, founder and CEO of networking company Aryaka, about these topics.

"We are at a point where cloud has reached a tipping point," said Gupta. "Cloud adoption is rapidly accelerating, and today, 47 percent of all traffic on the wide area network (WAN) is cloud-related."

Gupta said that many corporate IT departments are hard pressed to solve the performance issues that accompany a rapid increase in WAN traffic. "What they're finding is that traditional approaches like the virtual private network (VPN) still don't solve issues like bandwidth constraints," said Gupta. "To solve bandwidth for many companies becomes a cost-prohibitive investment, so they must find a way to accelerate their WAN traffic, whether that traffic has to flow through a straw or a large hole."

Gupta's organization focuses on this problem by providing intelligent network optimization for the WAN. Aryaka has a worldwide network of distributed points of presence (POPs) that is positioned near end users in major metro areas around the world. These POPs are then linked into a private network where resource availability can be monitored and optimized so as to produce predictable enterprise-grade global connectivity.

IoT end point data collection and analytics are central to the process. This enables a corporate network administrator to obtain WAN analytics that his own instrumentation can't provide, and to fuse these analytics with his own internal network data and metrics for a starting point on the road to end-to-end application monitoring. Being able to consolidate internal and external network analytics into an application performance composite also eases the minds of network administrators in their quests for end-to-end app visibility, because they realize it isn't necessary to "go it alone" for every element of big data harvesting and harnessing that ultimately factors into quality of service.

Gupta stressed the need for network administrators to take this broad view of application performance, given the fact that most companies now operate globally.

"In geographies like India and China, there are huge packet loss problems that bring down network performance and response times," he said. "This impacts the entire world economy because when you have many people trying to access a lot of data, businesses have to meet that demand to maintain a competitive edge. In the BRIC (Brasil, Russia, India, China) countries, there are also network issues because additional infrastructure must be built to handle traffic demands." These "last mile" Internet problems can be added to the U.S.'s own "last mile" problem, since large sections of the country still have inadequate Internet access and infrastructure.

The bottom line 

Net admins now have methodologies available that will aid them in facilitating end-to-end visibility of the network and application performance—because they can run and acquire analytics from both LANs and WANs, consolidate this data, and obtain information about how applications are serving users in different areas of the world.

This is big data and analytics in action--and it is also the "real" and "underlying call" of application performance management (APM) today. Now is the time for network professionals to respond to it.