Unlock the Value of IoT with Performance Monitoring

From cars to home appliances to wearables, the Internet of Things (IoT) has rapidly found its way into everyday objects, changing the way we live and work. In fact, IDC estimates in its Worldwide Semiannual Internet of Things Spending Guide that by 2021, global IoT spending is expected to total nearly $1.4 trillion.

It’s easy to understand the “why” behind the rapid rise of IoT and why top analysts like Gartner and IDC predict its continued expansion: IoT enables us to simplify our lives and be more productive. But what impact will the continued growth of IoT have on IT teams? What new challenges does it present? And, perhaps most importantly, how can IT teams successfully overcome these challenges to get the most out of their IoT initiatives?

A new white paper by MIT Technology Review delves into these questions, highlighting how the deluge of data created by complex IoT ecosystems creates a major obstacle for IT leaders looking to extract maximum business value from IoT. But, it also reveals how—with real-time performance monitoring tools that incorporate advanced diagnostics and data correlation capabilities—IT teams can surmount this challenge.

Here is a short summary of why application performance monitoring (APM) is the key to IoT deployment success, which IT pros can learn more about in this insightful new report:

The Challenge

As an emerging technology, IoT is disruptive. The size, scope, complexity, and volume of connected devices far exceed the ability of even the most expert IT departments to independently manage and oversee them. As Andrew Baker, CEO and Founder of Brainwave Consulting Company notes, “The explosion of devices and people trying to access and transmit information across IoT ecosystems is staggering, even for small firms with 50 employees.”

IoT infrastructures contain technologies with more complex applications, more robust server hardware, new protocols, new security requirements, faster and more efficient wide-area transmission mechanisms, and more backup procedures.

All of these new technologies tax the skill sets of IT departments and operations staff. IoT also generates much higher volumes of network traffic, making routine network management and capacity planning tasks more daunting.

There is just no way for businesses to manually keep pace with all of these connections, even if they have fully trained IT departments adept at IoT—there is simply too much data.

The Solution

Organizations can mitigate and minimize the impact of this disruption by deploying IoT advanced performance monitoring and management tools that deliver diagnostics and solutions in real time. These tools ensure the smooth configuration, provisioning, deployment, and ongoing maintenance of IoT-connected devices.

But knowing how to best deploy these tools is also a key step in getting the most from IoT—and creating a more nimble organization. When effectively implemented, advanced APM solutions can alert businesses to potential problems, enabling them to proactively address issues in advance of a problem that could result in a disastrous component or network outage. The business intelligence and predictive analytics capabilities of APM solutions can also identify trouble spots and avert service interruptions.

The Benefits

The top two benefits of application performance monitoring tools for IoT deployments, as revealed by MIT Technology Review’s research, are improved performance and security. In fact, 86 percent of survey respondents said the ability to quickly identify performance issues is a major benefit of APM for IoT, while 81 percent said the the ability to quickly identify security issues is a major advantage.

Effectively utilizing IoT monitoring and performance management tools delivers value in real time. Instead of taking hours or even days to identify a performance issue, the monitoring tool can quickly and efficiently locate the problem’s source, correlate the data, and enable the business to swiftly implement a fix. This improves reliability, increases customer satisfaction and retention, and lowers overall product support and ongoing maintenance costs.

With AppDynamics APM, for example, software teams can troubleshoot and quickly identify the source of problems so they can take fast, remedial action and lower mean time to recovery (MTTR) and repair to minimize or avoid costly service outages.

One real-world case study highlighted in the white paper is that of Nasdaq—the single largest U.S. stock exchange by volume. With AppDynamics APM, Nasdaq has dramatically improved visibility and time to resolution. The stock exchange has also gained the potential to automatically access new levels of actionable data. Nasdaq administrators, for instance, no longer need to perform the time-consuming task of scouring event logs to find issues. This has enabled them to cut down the time to pinpoint and resolve problems from hours or days to mere minutes.

APM is critical for identifying and resolving IoT performance issues, security problems, and outages much more quickly—and, ultimately, to getting the most out of IoT.

The Path Forward

IoT is a disruptive technology, but that does not mean the cost efficiencies, economies of scale, and revenue gains it provides will be automatic.

The true value of IoT lies in how effectively corporate enterprises can utilize and leverage it to achieve business goals such as cutting costs, increasing revenue, and optimizing labor resources. And, as MIT Technology Review’s new report highlights, advanced IoT performance monitoring tools are the key to helping IT teams unlock the full business value of IoT.

Read the Full Findings Report

To explore MIT Technology Review’s full findings on this topic, download the free white paper here. The white paper offers deeper insights into the rise of IoT and analysts’ predictions for the future, why IoT projects fail—and how to avoid them, and how APM can be used to drive IoT deployment success.

IoT and the Age of the User

It’s come to be the number to guide us through the IoT revolution, and all the tech innovations driven by it: a projected 50 billion connected devices by 2020. Forbes wrote of retailers betting big on the Internet of Things, changes have been made to accommodate the impending wave of data and new protocols, and governmental programs were drafted to try to regulate the deluge – but the storm never came as expected.

In fact, it looks like IoT adoption, although far from changing direction, has stalled considerably. Statista.com, for example, shows a steady growth but no geometric progression and puts the 2020s at the 30 billion mark. Gartner, too, revised their estimate to a much more modest 20 billion.

It may be that we’ve been wrongfully assuming that the biggest challenge for IoT evolution is connectivity – it’s not. Broadband is currently affordable in 111 countries, with a basic fixed or mobile plan costing less than 5 percent of Gross National Income (GNI) per capita. With that out of the way in developed countries, some turned, instead, to the other components: the IoT layers. From the device itself to embedded systems and from the data processors to the cloud, all grounds were covered in the past year by new technologies, services, and protocols. Still, not enough.

What are we missing?

The real challenge is perfecting the UI/UX layer by addressing the three pillars it sits on within every IoT deployment – safety, relevance, and performance – through a sleuth of improvements made possible only by continuous, deep monitoring.

Keeping the user safe and the notion of digital trust

Consumer devices are the main driver of IoT, with over 5 billion units sold in 2017 alone –  63 percent of the total. Smart cars, smart TVs, digital set-top boxes and wearables are as ubiquitous as cell phones nowadays, but with scandals like the Norwegian Consumer Council’s #ToyFail report and many, many stories of data breaches hitting the media, 62 percent of consumer IoT users are now very concerned with privacy, and 54 percent are worried about their data security. (Add to that the 21 percent who openly declare they’re scared of AI taking over the world, and the picture is quite apocalyptic).

The digital trust, before any app is installed, is an exercise of transparency. Failing to mention that you’ll record everything a kid says through their smart toy and then selling that info to third parties is a blatant breach, but so are other more sophisticated ways to collect information. As a result, smart users are getting smarter, and “what gateways does your railway use?” or “who’s your bank’s APM?” have a chance to become as meaningful as “Intel Inside.”

Keeping the user using and the notion of digital relevance

Customers procure, buy, sign up; but the truth is, ownership does not guarantee usership. In fact, at some point, of the 19 million registered Fitbit users, only 9.5 million were active. What’s more, about a third of owners of smart wearables ditch their devices or turn some of the smart services off. Even so, wearables are on the rise, with new launches announced each month, some of which are already making a huge impact – like Pebble and internal health trackers. Other lifestyle IoT are also here to stay. Smart Home applications are stable and rising as well, with the term itself, smart home, being searched over 600,000 times every day.

So why aren’t buyers using more? The explanation may lie in two things: First, the category is very fragmented. If data is shared with one app, it may skip another app or flatly not work well with another back-end software – and it’s simply too hard for the user to keep it all updated.

Secondly, faulty business logic or system limitations impede the automatic correlation of customer engagement data with business performance data (and vice versa). If the wrong consumer data is collected or looked at in the first place, turning it into insights and then action does not make a positive difference on the user side. In other words, the relationship with the app or service becomes irrelevant because the analytics are in the wrong and the offering no longer reflects a need.

The formula for making wearables stick is still a huge TBD and not much can be done independently to integrate platforms and apps from different providers. However, a lot can be done to affect one app’s usability and how it’s measured, thus making it more relevant—and competitive.

Successful companies not only cover these bases by employing the right kind of performance monitoring tools, but have discovered that the best monitoring is not a patchwork of disparate systems, but unique platforms to correlate real-time user behavior with business performance. Thus, getting real-time visibility can calibrate the right parameters, at the right time, all the time.

Keeping the consumer IoT usable with real-time performance

If digital trust and effective need are satisfied, the user experience relies entirely on performance, including speed of delivery, accuracy, and reliability.

That being said, IT teams now have to deal with several additional layers of complexity on top of the usual management conundrums: growing number of devices, variety of data formats, custom business logic, external threats, and the customary nature of IoT applications – all of these, and the relationships between them, leave monitoring with a lot of blind spots.

And with the sheer footprint of an IoT infrastructure (with all its layers) being as wide as it is, that means the lines between the integrity of the infrastructure, the performance of the infrastructure, and the performance of the user applications are blurred.

While automating each of these key IoT components is justified and desirable, they have to work together seamlessly to create a scalable, intelligent system—otherwise the puzzle is disjointed and unstable, and the full picture unclear.

So, again, what gives?

Speed of delivery (of services, analytics, and triggered actions) depends on the infrastructure and the performance management system in place. If the infrastructure supports stream processing capabilities and the system performance and application performance management (APM) tools are able to monitor, baseline, troubleshoot, and scale in real time, speed of delivery is optimal.

Accuracy determines relevance, and it depends largely on both the ability to monitor end user behavior, and—as previously mentioned—on correlating it with application performance levels and business outcome in real time.

Reliability, apart from previous indicators, is also conditioned by uptime. Continuous execution and the ability to detect failures and threats before they occur is paramount.

To help, AppDynamics IoT Monitoring offers real-time visibility, diagnostics, and analytics for IoT applications. To learn more, schedule a demo or start a free trial today.

Getting Ready for Connected Enterprises: How We Built AppDynamics IoT Monitoring Platform

The AppDynamics End User Monitoring product has certainly evolved over time. It first launched in 2013 to monitor web applications running on browsers. Then, to address the shift of increasing mobile activity (thanks in large part to the introduction of the iPhone), we added support in 2014 to monitor iOS and Android applications.

And now, as the Internet of Things (IoT) continues to grow, more and more user interactions and business transactions are originating from embedded smart devices.

To keep pace with this shift, we launched IoT Monitoring during our Winter Release to monitor application performance on any device connected to the internet. This includes applications running on connected cars, set-top boxes, industrial gateways, smart home devices, and more.

In our first post for this series, The Importance of Business and Performance KPIs for IoT Applications, we looked into the technical and business requirements for successfully deploying and managing an IoT application. In this blog post, we’ll dive into the details of how we built our IoT Monitoring platform and its use cases.

To start, we wanted our IoT Performance Monitoring solution to support:

– Ingestion of monitored data from all hardware platforms running on any operating system (e.g., embedded Linux, QNX, mbed OS, VxWorks) and application framework (e.g., C/C++, Java, Python, Javascript, Node.js).

– Capture and transmission of monitored data with minimal overhead to the application. The solution should also operate within the device constraints for memory, computing power, and network bandwidth.

– Performance monitoring of network protocols such as HTTPS, MQTT, and AMQP.

– Generic monitoring data model applicable to different IoT verticals such as retail, transport, media, and industrial gateways.

– End-to-end visibility starting from a connected device to a data center, network equipment, and all the way to the database.

With the considerations above, let’s look into different constructs built as part of the new AppDynamics IoT platform.

Data Ingestion

One of our goals for building the IoT monitoring platform is to enable any IoT device and application framework to ingest data to our platform. These devices could range from low-powered, limited-computing, micro-controller-based devices like a smart home, to high-powered, high-memory, microprocessor-based devices like connected cars and set-top boxes. To provide flexibility for developers to monitor applications written on these devices, we released a public HTTPS REST API along with lightweight C/C++ and Java SDKs.

The SDKs provide as much flexibility as using the REST API and also handle buffering, batching, and serializing data. They do not depend on any third-party network library but use the application’s network communication to send data. This provides the developer complete control on when to capture and transmit data to the AppDynamics SaaS platform. Sample applications showing the use of REST API and SDK’s are available on github.

Data Model

In IoT, the data generated differs by application and device type. As a result, the platform must be flexible enough to capture and visualize a variety of data.

For example, in point-of-sale devices, we want to capture payment data, items in the cart, and store information. On the other hand, with a media application running on a set-top box, we want to capture video streaming stats, number of active users, and ads displayed.

In both cases, data is sent to an IoT endpoint as a beacon in JSON format. Each beacon has four constructs as outlined below. An IoT application can send one beacon at a time or batch them and send multiple beacons up to the limit defined by the platform.


Metadata gives the platform context of the device and application configuration that is generating data. Two objects that capture metadata are DeviceInfo and VersionInfo. DeviceInfo contains fields such as device name, device type, and device ID, which help identify how many unique devices are reporting data and also derive device specific stats. VersionInfo contains fields such as software, hardware, and OS version, which help filter performance data based on the different versions.

Figure 1: Connected Devices Applications

In Figure 1, the Connected Devices tab provides a list of all applications and device types that each application is running on. You can see that there are three different applications and their respective device types:

– Retail Application has device types such as point of sale and smart shelves.

– Media Application has device types such as Roku, Fire TV, Apple TV, and Android TV.

– Car Infotainment Application has car models as device types, including Toyota, Honda Audi, and Mercedes.

IoT application data is grouped based on the device type as each device type has a unique profile in terms of the hardware, OS, and application framework. This data view enables businesses to quickly analyze application performance by device.

Figure 2:  Devices Dashboard

Figure 2 shows the Device Dashboard when Point of Sale devices are selected. The Device Dashboard presents a list of all the unique point of sale devices that were reporting data along with the metadata for each device.

Network Event

IoT is bringing connectivity to a lot of old and new physical devices. To provide a seamless user experience, it is important to ensure these devices are up and running with always-on network connectivity. Network performance is thus one of the key KPIs which help measure user experience.

AppDynamics Network Event helps capture the performance of any network request made by an IoT application. Currently, network event supports capturing HTTP requests and responses. In the future, we will extend it to support different network protocols prevalent in IoT, including MQTT and AMQP.

Figure 3: Network Request Dashboard

In Figure 3, you can see that the Network Dashboard provides details on network performance and all the URLs the application is triggering. It provides an aggregated view of network performance for the application on a specific device type.

Error Event

Reducing MTTR is a key objective for any operational team. So, it’s important to detect and diagnose application errors before it impacts user experience and business performance.

AppDynamics Error Event helps capture all types of errors including alerts, critical, or fatal errors. Alerts or critical errors are caught and gracefully handled by the application, whereas fatal errors can cause application reset.

Figure 4: Error Dashboard

Error Dashboards provide details on different types of errors, total error count, and error count grouped by application versions. Selecting a specific error will provide detailed information about the error, including stack trace if available.

Custom Event

Network and Error events help in understanding the performance of the application. But to understand the usage of an application and how that impacts business performance, we introduced Custom Events. These events help capture any data pertaining to the business, which can then be used to inform performance and business decisions.

Figure 5: Business Performance Data

As shown in Figure 5, using custom events for Point of Sale devices can help capture data such as total revenue generated, average sales over time, and the number of items sold.

End-to-End Visibility

One click or touch on an IoT device triggers a series of transactions across many components in the IT infrastructure. AppDynamics’ suite of products including IoT can tag and trace all the transactions across the entire infrastructure, thus providing end-to-end visibility.

Consider an example of self-serve movie kiosks where users can pay for their movie tickets. Figure 6, below, shows the journey of the transaction starting from the kiosk to the IT infrastructure.

Figure 6: Network Request Snapshot and Backend Business Transaction Correlation

Selecting the network request you want to trace shows an activity stream, and if the backend is instrumented with AppDynamics agents, you will see a snapshot link. The snapshot view shows how the business transaction is performing on the backend, and you can drill down into different KPIs for tier/nodes. This end-to-end visibility helps in tracking and identifying issues quickly, thereby reducing MTTR.

What’s Next?

In 4.4, we built a generic and scalable platform for IoT performance monitoring that can provide visibility into application and business performance for a wide variety of use cases ranging from point-of-sale devices to industrial gateways.

Looking ahead, there are a few emerging trends in the IoT space that we’ll be monitoring closely.

For starters, the Machina Research Annual Report predicts that the total number of IoT connections will grow from 6 billion to 27 billion by 2025. Of these, 71% of all connections will be using short range communication technologies such as wifi, Zigbee, or PLC. Adhering to this trend, there is an increasing adoption of new communication protocols such as MQTT, AMQP, XMPP, and COAP as they provide security and low overhead for IoT device communications.

What’s more, IDC analysts predict that the volume of worldwide digital data will be 163 zettabytes by 2025, and more than a quarter of this will be generated by IoT. Edge Computing is a paradigm that is playing an increasing role in better managing and deriving value from this volume of data. Edge Computing allows data to be processed near the source rather than sending it to the cloud or a data center. For example, security cameras such as Nest are using on-device vision processing to send alarms if it detects an unrecognizable person. Similarly, connected cars, smart cities, manufacturing plants, and building management systems are using data generated from sensors to derive time-critical decisions locally, instead of transferring data to the cloud and waiting for the decision.

To address these emerging trends, our IoT monitoring team is continuously evolving the platform to measure performance metrics of the new IoT communication protocols and also provide real-time performance insights at the edge, using local data filtering, processing, and modeling before sending it to the cloud. These metrics and insights will help businesses effectively manage the complexity and services in the space of IoT.

Learn more about AppDynamics IoT Monitoring by scheduling a demo or starting a free trial today.


The AppD Approach: IoT and AWS Greengrass

As both data and processing power rise on the edge of the network, monitoring the performance of edge devices becomes increasingly important. In addition to deploying the AppDynamics IoT monitoring platform to monitor C/C++ and Java apps, end-to-end visibility can be extended to applications running in an AWS Greengrass core by using AppDynamics IoT RESTFul APIs. The easiest way to do this today is with a Lambda function. We recently demonstrated this at AWS re:Invent using Cisco IOx and Cisco Kinetic together with AWS Greengrass on a Cisco Industrial Integrated Services router.

The best thing about this approach is that it opens up a new ecosystem of edge applications to the benefits of unified application monitoring. It ensures customers will resolve incidents faster, reduce downtime, and lower operations’ costs. Meanwhile, the combined strengths of AWS Greengrass and AppDynamics’ IoT Monitoring Platform allow very large volumes of data generated by the Internet of Things to be mined for business insights and harnessed to achieve business objectives.

AWS Greengrass is designed to simplify the implementation of local processing on edge devices. A software runtime, it lets companies execute compute, messaging, data caching, sync, and machine learning (ML) inference instructions even when connectivity to the cloud is temporarily unavailable. Since its release, it has helped accelerate adoption of IoT by making it easier for developers to create and test applications in the cloud using their programming language of choice and then deploy the apps to the edge.

Once the apps are deployed, AppDynamics’ IoT Monitoring Platform provides deep visibility, in real-time, by letting developers capture application performance data, errors and exceptions, and business data. Since the AppDynamics solution is designed for flexible integration at the edge, Lambda functions can be individually instrumented, or a dedicated Lambda function can be written to provide insight into all the Lambdas running. This allows for a wide range of edge applications to monitor any key metric that makes sense to the business.

In the demo at AWS re:Invent, we instrumented an edge application running on a manufacturing floor that was reading sensor data from a programmable logic controller (PLC) over a Modbus interface and reporting it back to the cloud. A key success metric was how edge computing reduced the large amount of inbound data volume to a much smaller meaningful volume that was being pushed to the cloud. AppDynamics provided real-time verification by keeping track of the volume of data being ingested into the Lambda functions, and of the data that was being processed and being sent to the various cloud applications, including AWS Cloud.

Learn more about AppDynamics IoT monitoring and please send us any feedback or questions.

The Importance of Business and Performance KPIs for IoT Applications

Over the last few years, the Internet of things (IoT) has become a trending phrase for consumers and a top priority for businesses embarking on their digital transformation. Even with the growth and interest in IoT however, the meaning can still confuse people.

So, what is IoT? IoT is a network of things connected to the internet and is uniquely identifiable through its embedded computing system. These “things” may include a variety of devices like home appliances, commercial vending machines, fitness trackers, industrial gateways, connected cars, and smart factories.

And worldwide spending on IoT devices is on the rise, with IDC’s Worldwide Semiannual Internet of Things Spending Guide predicting that global spending in IoT will leap from over $800 billion in 2017 to $1.4 trillion by 2021. This increase is attributed to continued investments made by organizations in the hardware, software, services, and connectivity that enables IoT. The goal of these IoT investments? To drive operational efficiency and increase revenue through improved consumer experience.

For example, the transportation industry is using sensors to improve fuel usage in planes and trucks, while the industrial sector is using IoT to reduce gas leaks. Environmental sensors for humidity, CO2, and electricity sensors also help reduce energy costs for buildings.

On the other hand, IoT in sectors like retail, automotive, and media are more focused on providing consumers with a rich experience by enabling new device interactions and avenues to consume data. For example, the retail industry is using devices such as smart shelves, point of sale, and digital signage to significantly improve consumer experience in brick-and-mortar stores to drive more sales. There are also voice-controlled devices like the Amazon Echo and Google Home, which offer a premium experience by allowing consumers to play music, stream podcasts, provide weather updates, control your smart home, and more.

And it’s these type of consumer experiences that drive sales and customer loyalty. In fact, IDC reports that consumer IoT spending will be the fourth largest market segment in 2017 at $62 billion, and will jump to the third largest segment come 2021.

Monitoring IoT Performance

As the number of IoT devices in the consumer and business space increase, as will the complexity of the infrastructures needed to support the new services and touch points. With this increasing software complexity, there is also a correlated demand from users for highly-responsive, real-time digital services.

As a result, a toolset to measure and deliver an exceptional end-user experience is imperative for making an IoT application successful. And that’s precisely where AppDynamics can help. AppDynamics IoT monitoring provides visibility into your connected device applications for real-time performance diagnostics and usage analytics so you can quickly understand and resolve performance issues.


Fig 1: AppDynamics End-to-End Performance Monitoring

In Figure 1 above, you can see how AppDynamics’ end-to-end unified monitoring solution provides visibility into a complex software infrastructure. AppDynamics follows the transaction at each hop, starting from a connected device to a data center, network equipment, and all the way to the database.

AppDynamics End-User Monitoring provides great visibility into browser and mobile applications and now – with our Winter Release – we are extending it to monitor all connected devices.

IoT Monitoring Requirements

Before we built our IoT Monitoring Platform to help operations teams manage IoT applications efficiently, it was important for us to understand monitoring requirements from both the technical and business end. We built our platform with the below technical and business requirements in mind.

Technical Requirements

– Ability to monitor IoT applications that run on devices with different processor architectures (e.g., ARM7, x86, Cortex-M series), and a multitude of operating systems. (e.g., embedded Linux, QNX, mbed OS, VxWorks)

– Ability to monitor IoT applications written in multiple languages (e.g., C, C++, Java, Python, Javascript, Node.js).

– Overhead for monitoring IoT applications should be minimal and operate within device constraints such as memory, computing resource, and network connectivity.

– Ability to ingest data generated by IoT applications that can vary significantly based on application type. For example, an industrial gateway device might generate gigabytes of sensor data whereas a point-of-sale device may trigger thousands of user transactions per day.

Business Requirements

– Ability to manage the complexity of software and services offered on the new IoT device types and applications. IT needs to detect issues proactively and keep MTTR low.

– Ability to provide the same user experience, independent of device type.

– IoT devices generate tremendous amounts of data and it’s important to be able to get insights into the business performance quickly.

– Ability to correlate business performance with IoT application performance. For example, when a business is losing money, it should be easy to quickly identify the root cause of a performance issue.

– Ability to react to real-time alerts on application or business performance issues.

Stay tuned for the next blog post in this series, where we’ll dive into the technical details of AppDynamics’ IoT product offering, how we solved design challenges, and how we’re helping businesses tackle IoT proliferation.

Learn more about IoT Monitoring or schedule a demo of our product today.

Conversational Technology: Siri, Alexa, Cortana, and the Google Assistant

One of the big laughs from Star Trek IV: The Voyage Home, where the crew from the future travels back in time to 1986 San Francisco, happens when Scotty is put in front of a PC and the chief engineer of the Starship Enterprise speaks into the mouse, “Hello, computer.”

In the 30+ years since that film came out, society has undergone a digital transformation. Now the average half-asleep user can roll over in bed and call out from under a pillow, “Hey Siri, start the coffee.”

The dawn of the Internet of Things (IoT) brings your daily environment alive with smart home and smart office devices. However, it’s up to innovative app developers to introduce creative controls for these devices. Here’s a look at the tools developers are using to integrate their software-defined innovations with the big four conversational platforms: Siri from Apple, Alexa from Amazon, Cortana from Microsoft, and the Google Assistant from Alphabet.

The Top 4 Virtual Assistants


In 2010, Siri was one of the first functional virtual assistants, quickly acquired and branded as an essential Apple function. Siri had been built on the shell of DARPA’s Cognitive Assistant that Learns and Organizes, or CALO. In Scandinavian languages, Siri means “beautiful victory,” and in Swahili it means “mystery” — but both could be nods at its origin in the Department of Defense. More practically, Siri’s name is also derived from its parent organization, SRI International.

Apple’s SiriKit helps developers integrate their apps into iOS and the early experiments in smart home infrastructure. Siri handles the voice-based user interactions and natural language recognition functions. It connects the app with the camera, ApplePay, VoIP calling, and other related services.


When Amazon was developing the Echo device, it needed a wake-up word that wasn’t likely to be confused with other mentions. Tests with the name “Amazon” tended to generate responses whenever a commercial for Amazon came on TV and would start buying items at random. Alexa was chosen for the virtual assistant when it was released in 2014, but users had the option to change its name to Echo or Amazon.

Amazon has created an Alexa Skills Kit, a collection of self-service APIs, tools, documentation, and code samples for developers who want to build onto the Alexa platform. They also offer a Smart Home Skill API for streamlined control of smart home technologies like cloud-controlled lighting systems or rooms with variable thermostat settings. This code is designed to run in the cloud, not on the Echo or the user’s mobile device.


The same year Amazon put Alexa on the market, Microsoft demonstrated Cortana at the Build Developer Conference. In 2015, Cortana was included as part of the operating system in the Windows 10 desktop and mobile platform update. This year, Cortana will be integrated with Microsoft’s Skype. Microsoft is trying to catch up with other conversational platforms through superior personalization. Cortana’s Program Manager, Marcus Ash, said, “Millions of years of evolution tell us that relationship means personification. If you don’t put a face on it and make it emotional to people, it’s just hard to believe that people will tell us information that will make Cortana really great for them.”

Microsoft has an advantage with decades of developer research to build a comprehensive software kit. Integrating Cortana in Your Apps at the Microsoft Virtual Academy has videos, slide presentations, a free trial of Visual Studio, and various other resources for developers. It even goes into testing procedures and error handling. Unlike Siri, which only works with Apple’s iOS apps, Cortana works with common applications like Hulu Plus and Facebook.

Google Assistant

When Google, Inc. reorganized itself under a parent organization called Alphabet in 2015, it freed up the Google brand to redefine itself. While Alphabet goes on to invest in wild ideas, from humanoid robots to extending human life spans, Google is turning into a virtual assistant. In direct opposition to the other three platforms, Google’s assistant doesn’t have a female name or the illusion of a personality behind the algorithm. Google’s Jonathan Jarvis explained, “We always wanted to make it feel like you were the agent, and it was more like a superpower that you had and a tool that you used. If you create this personified assistant, that feels like a different relationship.”

The Actions on Google program has a single API that can support various Android devices including phones, tablets, and watches. It includes both Direct Actions, where the user asks for something specific, and Conversational Actions, where the user asks for something general and the Google Assistant engages in a conversation to gather all the details. Google demonstrated the Conversational Action with an Uber request, where the Assistant moderated a conversation with Uber about where the user is going and which kind of Uber service they want to take them there.

Converging Trends Driving Conversation Platforms

Though most of this tech has been available for years, 2017 should prove to be an exciting year in conversational platforms as these five trends converge:

1. Mobile Natives

The success of chat apps like Facebook Messenger, WhatsApp, and Kik are driven by an entire generation of mobile-native consumers, comfortable with messaging and interacting with their devices.

2. Language Recognition

Chatbot personalities and AI natural language processing are far more accurate than ever in understanding speech and context-aware requests.

3. Mass Personalization

Sophistication in sensors and wearable tech have made mass personalization and predictive assistance both possible and profitable.

4. Secure Online Payments

Blockchain and related online payment tech have integrated themselves seamlessly into messaging and third-party apps.

5. Always-On Interface

Notification intelligence has advanced to the point where it can reliably serve as an always-on interface layer across devices.

Best Applications for Smart Home/Smart Office

Some of the most useful applications available to users without programming experience involve Alexa recipes on If This Then That (IFTTT.com). These should inspire developers to go further.

  • Phone Finder: This connects to the iPhone Finder on iOS, but is limited if the ringer is off. For Android, you can combine recipes to turn up the volume and play music.
  • Lock the Doors: At night before they go to bed, users can direct Alex to make sure all the doors are locked.
  • Change the Lights: To make sure you get the message when a timer goes off, even if you don’t hear the alarm or are outside, Alexa can change the lighting to make sure you see it.

In the office, Microsoft projects that Cortana will be able to streamline your day at the office and make everyone more productive by discovering where time is wasted. For example, you will be able to ask Cortana who is in the office and when they are out of a meeting. Cortana will also be able to help assemble teams for collaborations, discovering who has the best skills and experience needed for each project. It will do this by culling information from emails, company documents, web searches, and software authorizations.

The big news from Google is their Google Cloud Speech API, which will be set up to handle more than 80 languages. The intent is to integrate with any application in real-time, streaming or batch mode. The API will be able to support communications from any device that can send REST or gRPC requests, including phones, PCs, cars, TVs, and other IoT devices. Google claims that this tech will be able to identify and respond to commands even in noisy environments, which could open it up to other types of work environments and public areas. Initially the API will be free, but Google plans to add tiers of paid service levels later on in development.

The future of Apple’s Siri is likely to include a standalone device to compete with Amazon Echo. Apple suggests that Siri will be able to search through images and videos to find either personal photos or an applicable YouTube video. Siri also demonstrated an ability to understand vaguely worded commands like, “Tell Nancy I’ll be five minutes late with WeChat.” A vision of what’s possible is the MapMyRun app from Under Armour. The next version will allow users to tell Siri to start, pause, and resume a workout among other functions. The built-in brand loyalty of users on iOS, OSX, and Apple HomeKit makes Siri’s growing openness particularly attractive for app developers.

The Default Interface for the IoT

Virtual assistants will have to be the average user’s primary interface with the IoT, at home or in the office. There will be at least 28 billion IoT devices by 2020, so knowing how to integrate your applications with conversational tech will be a critical skill set in the very near future. As people grow more dependent on these assistants, you’ll need more sophisticated application performance monitoring (APM) software for continuous delivery. Maximize uptime and use dynamic baselining to predict abnormal behavior patterns before they compromise performance. AppDynamics is positioned to handle what’s coming next, so make sure your innovative creations don’t get left behind.

Learn More

Learn more about IoT with this eBook, “Breaking Down the Internet of Things.”

Inside the Internet of Things at AppSphere 2016

I always look forward to seeing what’s in store for emerging technology at AppSphere. It’s been an exciting year already for all things connected, from wearables and sensors, to the enterprise ecosystem around the Internet of Things–and it’s only growing. Today, every opportunity around the Internet of Things represents the potential for exponential growth. In IDC’s Global IoT Decision Maker Survey, research found that 31.4% of companies are already working with an Internet of Things initiative, with an emphasis on cloud computing and analytics, and an additional 43% said they are looking to deploy their projects in the next year. The potential value of IoT is clear now; over 55% of those surveyed believed that an investment in IoT and emerging technology would give them market leverage.

It’s clear the value of the Internet of Things is growing with opportunity, and it’s critical for businesses adopting its technology to secure the right solutions to deliver the ideal experience to their customers. This year at AppSphere, we’re getting the best resources to share best practices, technical deep dives, and hands-on education to deliver everything you need to know on making the most of the IoT landscape in your enterprise.

Take a look at some of these sessions, and even more on the AppSphere session catalog you’ll have the chance to see live in a few weeks!

The Internet of Things in the Enterprise: Why your Monitoring Strategy Should Include Connected Devices

Amazon Echo and Google Nest have changed our lives as consumers. As the Internet of Things (IoT) grows, it can disrupt traditional business models. When connected devices start communicating with applications, the end user experience can be severely impacted. This talk highlights some of the unique challenges that monitoring millions of IoT devices will bring. It also emphasizes the features of a good IoT-ready performance monitoring solution.

Since IoT initiatives are new for many businesses, many have not completely explored what supporting them will mean to their monitoring strategies. Coping with the new ways users interact with applications across multiple platforms represents a unique challenge. In order to tackle the IoT obstacles, it’s increasingly necessary to develop a monitoring strategy that considers connected devices.

  • How connected devices impact the bottom line of your business
  • Why connected devices need a monitoring strategy
  • Best practices for monitoring connected devices

You already know about all the ways the IoT influences the user experience, but with about 43% of companies adopting IoT by the end of 2016, you may need to face the challenges of monitoring those user experiences sooner than you think. Join us for this session on Tuesday, November 15 at 2:30 PM.

Customer-Centric Transformation in the IoT Age: A Panel of Enterprise Innovators

Join us for a lively panel discussion with technology leaders from major global brands on how cloud, mobile, and connected applications are fundamentally changing their engagement with customers.

This panel will explore the new trends in IoT from leaders in the field and demonstrate how a customer-centric approach is necessary to navigate the technology trends. Not until recently were we able to hail an Uber or control the thermostat of our homes with just the sound of our voice. Technology companies are setting the tone of how users expect to interact with applications, so businesses need to keep up with the trends or risk being left behind. Come join innovators in IoT and get their perspective on how being customer-centric in their design has allowed them to capitalize on IoT trends. You can find this session on Tuesday, November 15 at 3:40 PM.

Stay tuned! We’re bringing you more sneak peeks on these sessions and even more in the weeks to AppSphere. In the meantime, make sure you’ve registered–spots are filling fast!

Why Analytics are Essential to the Internet of Things

In previous posts, we’ve touched on the key values of the Internet of Things (IoT) and the criteria to check before adopting into IoT. Despite mixed messages on security and standards, there is a clear takeaway that a tidal wave of data is on the way. The major question remains–how do we convert all that data into useful information?

The volume and speed of this connected data will overwhelm organizations that aren’t able to mine quickly. Analytics act as the translation systems for IoT data. The careful application of analytics can unlock an endless wealth of information in support of rapid decision-making amid volatile economic indicators. Resting on top of a sea of structured data, your business will acquire a clear vision for strategies such as: how to tailor your best offerings for preferred customer segments, how to streamline your internal processes for a lower total cost of ownership and how to develop the items or services that are already in demand in the market.

Related reading: Breaking Down the Internet of Things

Optimizing your IoT Performance

Analytics can help businesses capture the maximum value from connected data, but they have to be prepared before the flood of data starts to arrive. There are four core requirements for accomplishing this:

  1. The ability for the solution to store a high volume of data economically.

  2. Automatic correlation of performance data with customer engagement.

  3. The ability to allow managers to easily query the data in order to turn it into information and insights.

  4. Make the process repeatable and scalable to support any IoT deployment in the future.

Developing IoT for the End User

Amid the discussion of how businesses need to adapt to IoT, it’s easy to forget that IoT can still be a baffling change for the average consumer. To business users and developers who have seen the steady rise of M2M communication and intelligence, IoT is just the next step. Consumers with their attention locked on other priorities often express that they have no time to learn new interfaces. An IoT based product/service should be invisible to the consumer. As Steve Jobs famously mused, new technology should be as effortless as a refrigerator. That’s one of the reasons that he introduced the iPod with a single button.

Due to the fact that IoT is often being built directly into familiar devices, it has to go further. Ideally, IoT should sit in the background, enhancing experience and their daily life, like the new smart thermometers. To get there, most companies will need to use analytics to further refine their IoT services/products until it reaches that point.

In outlining the phenomena, Alvin Toffler said: “We may define future shock as the distress, both physical and psychological, that arises from an overload of the human organism’s physical adaptive systems and its decision-making processes.” Individual IoT gadgets (such as smart thermostats, drone-mounted cameras, real-time inventory sensors and health wearables) improve the end user’s life, but the cumulative effect can trigger IoT overload.

The same applies to all of your customers, whether that means consumers, businesses or end-users. The solution begins with seeing the end-product from the customer’s point of view and how it fits into their lives. In other words, the answer starts with analytics.

The State of Analytics

The state of the current analytics market is that the offerings are not suited to the needs of IoT. The main types of analytics available are:

  • Business Data Analytics – Concerned with metrics like revenue, new customer growth, and churn. Examples include Oracle Business Activity Monitoring, Teradata, and GoodData. These are too siloed to capture all the ways IoT will impact your business.

  • Marketing Data Analytics – Concerned with customer preferences, behavioral information, and new feature usage. Examples include Google Analytics, Omniture from Adobe and Webtrends. These are too narrowly focused and require specialized skills to interpret the results.

  • Application Analytics – The unified solution that focuses on a real-time analysis and visualization of automatically collected data to get insights into IT operations, customer success, and business outcomes. Leverage Transaction Analytics to deliver answers on performance and business metrics in real-time, User Analytics for a comprehensive IT operations visibility, and Browser & Mobile Analytics to optimize your end user’s journey.

It takes a more unified analytics solution to educate and develop the end-user to the point where the most effective, personalized IoT solution feels essential to their productive lives without triggering a future shock reaction. AppDynamics Unified Analytics actually connects the dots between your application’s performance, end users and business outcomes in real time. Auto-correlated rich, integrated data not only optimizes customer experiences, but drives a better business outcome as well. With the next generation of Unified Analytics solution – enterprise customers can quickly answer more meaningful questions than ever before, all in real-time, to power their connected initiatives.

Responding Faster to the Connected World

The future demands that you see, act and know in one elegant motion. Like all technological solutions, that sounds simple on the outside but is enormously complex on the inside. You will need instant answers to questions that cross analytical boundaries, such as:

  • What was the total value of Gold-level customers purchasing new features?
  • What is the breakdown of who is driving our cars at the moment, by customer tier?
  • Which device is most popular for connecting to our customer app?
  • How many new customers have signed up for our service?


It can be incredibly frustrating for everyone concerned when basic management questions like these demand an intense amount of IT resources – but that is the future. Options in the world of analytics must evolve to meet the challenges of IoT and the new market forces.

We’ve built a more evolved solution to specifically address those forces. We streamline the entire IoT development process by giving you one unified platform to work on, with a single install and a consistent UI. Our production monitoring requires a very low (<2%) resource overhead. You can jump right into processes because we can auto-discover complex transaction flows and save you the hours you’d normally spend on manual configurations. Just like the Internet of Things itself, we set a baseline of healthy performance and learn from there, alerting you immediately to any performance deviations.


IoT applications cannot operate without a more comprehensive understanding of your customers and what they are striving to achieve. The right analytics will deliver reliable performance data on the complex mix of software, hardware, networks and third parties that make up any IoT application.

A flawless performance is the heart of your business. Your analytics should isolate any mission-critical issues and guide you in resolving them long before your customer service department hears about it. Adding the right analytics module to your connected applications not only brings real-time analysis and visualization of automatically correlated data to get insights into cross-functional outcomes, but also enables IT and business users alike to quickly answer more meaningful questions than ever before, all in real-time.

The Key Values of the Internet of Things [eBook]

A few weeks ago I talked about the criteria your team needs to assess before moving forward with adopting into the Internet of Things (IoT) for enterprise. There is definitely no shortage of resources when it comes to accessing the Internet of Things in the market. In fact, the venture capital funding for IoT startups totaled $7.4 billion over six years. However, that does not mean creating an IoT business is easy–your company needs to overcome multiple hurdles in order to create successful IoT devices. We recapped some of the most common challenges you might face in IoT, and how to tackle them.

1. Devices

The challenges within the Internet of Things go beyond making devices that work; both product and service need to work seamlessly, almost invisibly to an end user. This is true for both the IoT and the consumer side. The service needs to meet needs, easily integrate into daily life or the industrial process, and has to enhance the user’s life or the business process. That means IoT solutions can’t focus simply on reliable hardware; the software backends need to be reliable as well. Consumers expect both their device and its website or app to function 24/7. When devices control real world environments, they cannot simply shut down due to a software bug, as some users experienced with the Nest thermostat. For industrial applications, the consequences of a device shutdown are far more severe than sleeping under an extra blanket. Security poses another challenge. There have already been worms targeting connected devices, such as security cameras; researchers demonstrated they could take over a driverless car. Security experts expect “machine to machine” attacks to increase during 2016. In fact, the risks are so real that the FBI issued a public service announcement to alert companies and the public to the dangers.

Companies also need to develop a strategy to cope with all the data collected by IoT devices. The volume of data is enormous; a jet engine generates 1TB of data every flight. Companies will need to cope with the massive amounts of data, combine data from multiple sources, and run efficient analytic processes. Finding ways to work with it may require adopting new techniques, like edge/fog computing, to reduce the amount of data sent to the backend systems.

2. Process

To build a level of software required by an IoT business model—with high capabilities of availability, security and performance—companies need to bring in skills they may not have needed when developing other products — embedded programming, real-time event processing, big data analytics. Before bringing that level of skill sets, it is important to think about the IoT product’s value to the user, it has to remove friction and perform a practical and useful service. For your product to succeed, the focus must return to the end user. You also need to be careful about committing to specific technologies. It is still early for many IoT standards, so expect to change your architecture as the technology evolves. There are multiple choices for almost every component of an IoT device — different chip vendors, various communication protocols, numerous backend software platforms. Future developments in sensors and batteries may someday let you implement additional functionality that isn’t possible right now. You may also need to adapt in order to scale your device and backend applications to larger volumes of data.

3. End Users

The primary challenge of the IoT device is that its design has to be customer-centric. End users won’t implement technology just for the sake of technology; the hype only goes so far to generate buzz. To achieve market loyalty, the technology must be able to simplify.  The devices also need to mesh with societal expectations. For example, the insurance issues around self-driving cars, including concerns such as the extent of the manufacturer’s liability, aren’t settled. Some insurers may want to adjust underwriting based on data about driving habits gathered by sensors in vehicles. This kind of usage of data raises privacy concerns the public may not be comfortable with.

Building an Ecosystem

Your IoT offering requires more than just a fancy hardware device. It needs to be part of an ecosystem in which the hardware is just one element. The ecosystem needs to include features for end users. This means a website or smartphone app where they can adjust device configuration settings and monitor device activity since things typically have neither displays nor input mechanisms to interface with.

The ecosystem needs to include features for third-party developers, to encourage its adoption by allowing others to create add-ons. This means creating APIs and tools for developers to use and implementing a process by which you can test their creations for safety before offering them for download in an online store.

Finally, the ecosystem needs to include features for your own developers. They need tools to monitor the usage of your product “in the wild” to keep an eye on performance and identify issues. Performance testing not only needs to make sure your production environment will handle the volume of messages received but also how your system operates if the connection fails and messages are not received. Your ecosystem also should provide a way to automatically deploy bug fixes and firmware updates to your products.

Outweighing the Challenges

There are inevitable complexities when it comes to creating and maintaining a consistent IoT product and software. With those challenges, however, come more opportunities for innovation and revenue. There are three main opportunities for companies to implement IoT business models:

  1. Digitize current processes or services: Tasks that are currently performed manually can be automated. For example, remote patient monitoring lets patients transmit vital signs to their doctors’ offices automatically, eliminating the need for follow-up visits.
  2. New business models: For some firms, using IoT devices can mean changing their business model. The World Economic Forum predicts a new “outcome economy” in which sensors will enable companies to charge for usage and guaranteed quality levels. Rolls-Royce, for example, charges per engine flying hour for its TotalCare aerospace service.
  3. Enhancing customer experience: Customers benefit from the always-connected aspect of IoT devices and being able to look at their data themselves. Companies can also take advantage of the data collected from the sensors to understand how customers use their devices and identify opportunities for new features and services that will make customers even happier in the future.

Your IoT offering requires more than just fancy hardware. It needs to be part of an ecosystem in which the hardware is just one element. The ecosystem needs to include features for end users and developers. Sometimes it means providing a front-end application for users to adjust device configuration settings and monitor device activity. Third-party developers must be encouraged to adopt by allowing them to create APIs and implementing a process by which you can test their creations before offering them for download in an online store.

Finally, the ecosystem needs to include features for your own developers. They need tools to monitor the usage of your product “in the wild” to keep an eye on performance and identify issues. Performance testing not only needs to make sure your production environment will handle the volume of messages received but also how your system operates if the connection fails and messages are not received. Your ecosystem, if built correctly, must provide a way to automatically deploy bug fixes and firmware updates to your products to ensure application and device performance is always providing the ideal experience to both developers and end users.

Get more out of the Internet of Things, and read the full eBook, Breaking Down the Internet of Things here.

4 Questions to Ask Before Adopting the Internet of Things

Stop me if you’ve heard this one before: there’s another side to the Internet that isn’t just about connecting people. The Internet of Things (IoT) is about connecting virtually any “thing” or machine. They could range from personal wearables to smart homes, smart cities’ infrastructure, utilities, transportation, and manufacturing. You’ve probably heard by now; the IoT is far bigger than the Internet of people, and it’s growing fast. Gartner says the IoT will grow 30 percent in 2016, reaching 6.4 billion devices, with more than five million new devices connected daily. It’s expected to continue growing to 20.8 billion devices by 2020.

It’s more than likely any business today sees the potential of adopting an Internet of Things model into their enterprise, but are you doing it in the best way? We set aside the four questions you and your team should be answering to determine how to find the right opportunity in the IoT space for your business.

1. What is the Internet of Things?

First things first. Defining the Internet of Things, and establishing its context to your framework is instrumental in determining your potential with IoT. Gartner defines the IoT as “the network of physical objects that contain embedded technology to communicate and sense or interact with their internal states or the external environment.” Verizon identifies three characteristics of IoT devices:

  • Aware: The devices include sensors that report information about their surroundings.

  • Autonomous: IoT devices are connected and automatically transfer information to a central location or application for processing.

  • Actionable: The information collected is integrated into business processes for decision making.

Figure 1: The three characteristics of IoT Devices

2. Is your software configured to connect with the IoT?

Talking about the IoT in terms of things makes it sound like the IoT is all about physical hardware. While the IoT doesn’t exist without sensor-based devices, the devices don’t actually function without software. Most IoT devices have a user interfaced website or smartphone app where the user can manage configuration settings and review activity. Some IoT devices have more sophisticated analytics that gather big data and crunch the numbers to make decisions about what the device should do or present insights to business management.

While building the software that runs on a device may require specialized skills for embedded programming, the backend processes are conventional software applications with common software development concerns, including performance and ease of use. The usual security concerns around software become even more important with the IoT, as software controls devices in the real world and security failures can impact physical systems.

3. Do your team’s technical capabilities scale with the needs of the IoT?

Technology developments now make dealing with the technical challenges of IoT devices easier. Platforms like Raspberry Pi provide low-cost boards equipped for IoT development. Low-power sensors and new low-power communication technology, such as LoRa, mean the limited power available to IoT devices does not limit functionality. Sensors and circuits are shrunk to the point that they fit into devices a person is willing to wear.

On the software side, companies have made platforms to create a standardized environment for IoT development. Applications can use RESTful APIs or lightweight protocol, which were designed to work where memory and network capacity are limited.

Both Amazon Web Services and the Google Cloud Platform offer features explicitly intended to meet the needs of IoT applications, including both real-time communications with IoT devices and performance of the big data analytics necessary to make sense of data once it accumulates. Combined with the hardware platforms, these services make it easy to get started prototyping a device and its software. Because prototyping platforms are scalable, if an idea is not successful, it is easy to continue developing it and create a robust product without throwing away the work that was already done. 

4. How does it create a ROI for your business?

Customer-generated data collected by IoT devices offer companies insight into customer behavior and create additional selling opportunities. They provide an in-depth insight that provides opportunity for companies to forecast everything from product roadmaps to market leverage. There are four main kinds of benefits for businesses:

  • Improvements in operational efficiency and asset utilization: Companies gain the ability to manage equipment remotely and schedule preventive maintenance to eliminate downtime. The IoT can also help with optimizing supply chains and loss prevention.

  • An outcome-based business model: The tracking and monitoring enabled by the IoT lets companies change the way they sell equipment. The use of sensors allows them to sell based on usage and quality level, allowing capital goods to adopt the “as-a-Service” model that’s become popular for software.

  • Analytics-based controls: Combining analytics with smart devices will let companies fine-tune control over their processes. Adjustments can be made in real time to ensure continued production and compliance with environmental standards.

  • Improved work efficiency: Smart devices will allow increased collaboration between workers and equipment, improving productivity.