APM Application Performance and the Internet of Things - Part 1
The challenges of managing performance across hyper-connected IoT systems
December 13, 2017

Gokul Sridharan
CA Technologies

The internet is full of predictions on how many devices will become Internet connected - according to GE for example who estimate 50 billion by 2030. Some studies claim even more, but perhaps what's more interesting is the business potential. According to McKinsey Consulting, the Internet of Things (IoT) will have a total potential economic impact of $3.9 trillion to $11.1 trillion a year by 2025.


But beyond the hype and cutesy examples about connected toasters and toothbrushes, there's no doubt that the burgeoning connected Interwebs will have a huge impact on every business vertical. Take for example:

Healthcare- where IoT is projected to reach $117 billion by 2020. And it's more than monitoring fitness wearables and health apps. IoT devices and applications carry huge societal benefits – for monitoring elderly people in their homes, with smart devices sending alerts to medical practitioners and family members in the event of problems. Helping better track high-risk patients, improve quality of life and reduce hospitalization costs.

Transportation– where IoT can play a proactive, real-time role in monitoring traffic conditions and providing analytics for better traffic management and routing. When integrated with GPS and messaging systems, these types of systems allow commuters to consider alternate routes when traffic become congested. It's also an invaluable tool in an urban planning context.

Automotive– where sensors intelligently monitor every aspect of a car to determine problems. Like with many applications, this data could be and analyzed in the cloud. Let's say for example a car sensor detects an impending deployment problem with safety airbags. Of course, the driver should be alerted, but integrated cloud-based applications ingest, correlate and analyze all the data – perhaps identifying that a certain class of vehicle at a certain level of mileage are exhibiting the problem.

Mining– where IoT devices can monitor the health excavation equipment – vital to ensure these capital-intensive assets are operate optimal condition and costly maintenance is kept to a minimum. In a safety context, IoT could be used to monitor mine safety conditions, with any breaches in safety thresholds triggering evacuations and other life-saving procedures.

Irrespective of vertical, the benefit of connecting "Things" – be that people, cars, homes, agriculture – whatever – is incredible. When businesses connect they unearth powerful use-cases that reduce operational costs, improve insights and can ultimately create new business models. And when what we connect becomes part of a more expansive IoT ecosystem the possibilities are left to our imagination. Like for example power management, where it's now routine for smart thermostat data to me mashed-up with utilities cloud data to predict high-demand and adjust settings to save cost – for both the consumer and energy provider.

Without doubt, smart devices can generate enormous amounts of data which together with analytics can drive unprecedented innovations; even new business models. But as IoT accelerates towards driving mission (even life-critical) applications, the need to manage performance at a massive scale increases incredibly. Failure to consider and address performance can have a dramatic impact on businesses.

So, considering the above scenarios, what happens when API and network latency problems mean healthcare responders are slow to detect a home alert? In the automotive use-case, what happens when analytics detects thousands of immediately at-risk cars, but interrupted back-end processing results in slow ordering and the transport of replacement safety air-bags? In any safety-related use case, what happens when a coding change to firmware results in an app performance condition that drains battery life – and that update is deployed to thousands of devices?

Managing mobile devices is child's-play compared to the challenges involved in monitoring performance across hyper-connected IoT systems. Depending on the use-case this can include:

Determining what should be monitored– by outlining the correct KPI's and performance thresholds. Metrics can be based on many factors – such as in-situ operational conditions, device availability and time-sensitivity. In more complex and interconnected system scenarios, the response between the sensor/device and back-end applications will be critical – especially where devices interact via APIs and gateways to core systems – including cloud analytics and Enterprise Resource Planning (ERP) applications.

Determining how and when "Things" should be monitored– devices will vary in processing power and capacity. In many cases this dictates low monitoring overhead because of device memory constraints. In other cases, device access for performance testing will be problematic and localized data storage constraints may dictate more advanced monitoring once information is transported to back-end systems. When performance needs to be established is also critical – when a device throw an alert or within and across connected applications.

Of course there are many other factors, not least managing performance across a variety of programming languages, operating systems, proprietary messaging protocols and emerging standards. Additionally, delivering scale is a huge factor since an IoT monitoring system could comprise literally millions of devices – delivering varied data types at different levels of speed and intensity. All of which needs to be ingested and correlated with other data sources and metrics across an interconnected fabric of applications, infrastructure and networks.

In future blogs, we'll zoom in on many of these challenges and present some essential application monitoring strategies. Some topics include:

■ Describing the new applications and distributed architectures underpinning IoT development and deployment and where application performance management should be applied for maximum effect

■ How to identify and mitigate against many monitoring constraints that are indicative of IoT applications.

■ Exploring ways to capture and combine IoT Big Data – correlating a variety of data sources, including operational and business metrics

■ The increasing importance of analytics and digital experience insight as a means to drive IoT business value

It's not a question of if IoT will impact your business, it's a matter of when. According the recent research, 2016 saw a 34% increase in focused on IoT. Their success will be wholly dependent on achieving new levels of application performance never before imagined or conceived.

Stay tuned for part two of this blog series where we'll be discussing IoT systems and where application performance monitoring presents itself as an important enabler.

Read Application Performance and the Internet of Things - Part 2

Share this