This vendor-written tech primer has been edited by Network World to eliminate product promotion, but readers should note it will likely favor the submitter’s approach.
Understanding how end-users access and experience network applications requires a new approach. Ten years ago, enterprises could count on Ethernet-connected, Intel-based Windows devices accessing on-premise applications. This made it possible to improve end-user experience within established use cases. But that scenario is the exception today rather than the norm.
Most applications today are supported by a dynamic infrastructure distributed across on-premise and the public cloud. And client devices are copiously diverse in terms of hardware, operating systems, drivers, and more. To add to the overall complexity, these devices are accessing networks via Wi-Fi over dynamic RF environments and potentially unreliable Internet connections.
What insight can traditional application performance monitoring (APM) and related technology provide in this new landscape? APM tools answer the question of how long a client transaction takes to get satisfied once it hits the web server. This takes into account the transaction times between the web server, app servers, database servers, etc. APM homes in on the inner workings of a multi-tier application within the data center. Even a brief consideration of contemporary applications, however, reveals that this portion now constitutes just a tiny segment of the overall user application experience.
First, IT is increasingly working with business critical applications beyond its control. When third-party applications are in use, IT cannot instrument a third party’s data center and the use of APM technology becomes moot. Second, using APM presupposes that clients are able to efficiently connect to the access network and subsequently reach the application webserver. When this connection fails, APM tools are not equipped to collect or analyze data from the wireless LAN infrastructure, the campus network services, nor the health of the WAN/Internet link.
Limited by its scope, APM is simply not the right tool to address the complexity of issues users are facing before their transactions even hit the application webserver.
What’s missing?
To gain maximum visibility into the user application experience, network operations staff need insight into the client device’s ability to associate in a timely manner to a wireless access point, authenticate via RADIUS, obtain an IP address via DHCP, resolve domain names via DNS and finally transmit/receive data from the internet. If any of these steps fail, it would be pointless to consider the “application performance” since users have failed to even access the application. Maintaining complete awareness of all of these user network transactions is daunting for network managers who must sift through volumes of network, application and user data from a variety of domain silos.
Evolving interdependences between networks, devices and applications are driving the need for a more holistic approach to IT analytics. Performance metrics about the application, wireless/RF, network services, Internet link, as well as device type and operating system information must be collected simultaneously across the stack. Next, these metrics need to be analyzed and correlated across time and other dimensions (e.g. locations, SSIDs, VLANs, etc.). Finally, advanced machine learning algorithms need to be applied on top of these metrics in order to surface insights and remediation advice proactively.
Network managers largely concern themselves with the operation of the wired and wireless infrastructure, but as cloud-based SAAS applications, smart devices, unified communications (UC) and Wi-Fi become more pervasive, a broader, client-based perspective is required to answer fundamental questions about end-user experiences.
User application analytics arrives
A more complete view of enterprise application behavior from the perspective of the user requires simultaneously collecting and correlating client wireless performance metrics, network service performance metrics, device/OS information and, finally, application performance metrics. The most frictionless way to accomplish this is with a combination of real-time packet analysis along with data collection from other enterprise systems, including the wireless LAN controller, as well as APIs from UC and other applications. Finally, advanced analysis including machine learning must be performed on the data in order to automatically surface insights.
New user application analytics (UAA)software technology looks to provide precisely this broad scope. UAA software runs passively within the enterprise access or campus network without the need for client software, synthetic testing, additional sensors or infrastructure hardware. Real-time application identification technology and deep packet inspection of wired packets, along with Wi-Fi metrics from wireless LAN controllers, are analyzed across the entire OSI stack for every client transaction as it occurs on the network. Applications are identified and performance metrics measured in real-time by looking at underlying protocol (e.g. HTTP, TCP, RTP, etc.) interactions and response times.
How user application analytics works
Application behavior for a given user, user group, discrete application, location or even a specific virtual wireless network (SSID) can then be determined and compared across a number of different dimensions to derive real answers to questions about the actual end-user application experience, such as:
- What is the baseline for user experience with application X?
- When the user experience of an application is poor, is it due to Wi-Fi, the device, the Internet link or the application itself?
- How does user experience vary across locations, subnets, SSIDs, RF bands, etc.
- Can users connect reliably to Wi-Fi and are link layer services being provided properly?
- If I increase the capacity of my DNS server, how many client-hours worth of SaaS user experience problems can I mitigate?
With UAA, IT staff are able to easily determine application adoption trends, traffic usage, individual and systemic client incidents and root-cause, historic trends as well as overall quality of experience for any user at any time.
Unified Communications monitoring presents one of the strongest use cases for user application analytics largely due to the prolific uptake in UC and the investments in digital transformation of multimedia business content.
For UC applications like Skype for Business, UAA hooks into vendor APIs to understand the user application experience with even greater fidelity. Call attributes such as mean opinion score (MOS), a standard telephony measure describing the actual user experience, are ingested, analyzed and correlated with other network transaction metrics.
Today's enterprise applications are becoming highly distributed, often with components spread across on-premises data centers and public clouds. For network managers, a good user application experience can no longer be divorced from the underlying wireless access infrastructure, network services and mobile devices used to access them. This now places IT staff in a precarious position of having to delve deep into volumes of network, client, and applications data to understand the actual user experience.
New user application analytics technology promises to finally give IT staff a complete view of their network from every angle and quantitative justification for the billions of dollars in IT investments being made every year to keep user productivity on enterprise networks humming.
Learn more by visiting Nyansa.com.
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.