Use of Predictive Analytics in Application Performance Management

What is Predictive Analytics?

Predictive analytics uses statistical techniques and algorithms to analyze historical data and forecast future events. Predictive analytics can benefit your IT teams in many ways. It gives the ability to monitor the health or status of an application so that you can predict and respond to application outages. The ability to prevent failures before they happen is a big win. It saves time and money and creates a more resilient IT infrastructure.

Predictive analytics tools can also be used to predict the user experience at a projected user load and evaluate whether the current infrastructure can support the projected user growth. They also aid in APM to boost overall application efficiency.

What is APM?

Application performance monitoring (APM) is a subset of Application Performance Management that focuses specifically on the real-time tracking and reporting of an application’s performance metrics. It helps monitor the user experience and leverages predictive analytics to improve application performance whenever there is a sudden spike in the number of users accessing the application or whenever there is a sudden degradation in application performance.

Predictive Analytics Usage

Organizations can leverage predictive analytics to improve their application performance in the following areas:

Identify root causes for application performance issues

By identifying root causes for application performance using machine learning techniques, organizations can focus on the right set of areas in which to take action. Predictive analytics can then study the characteristics of the various attributes within each cluster that can provide deep insights into what changes need to be made to achieve ideal performance and avoid specific bottlenecks.

Monitor application health in real–time

Performing real-time monitoring of application health via multi-variate machine-learning (ML) techniques allows organizations to catch and respond to the degradation of application health on time. Most applications rely on multiple services to capture the true health of the application. The data might consist of configuration data, application logs, network logs, error logs, performance logs, and more. Predictive analytics models can analyze past data during a time in which the application was in a good state and subsequently identify whether the incoming data exhibits normal behavior or not.

Predict user load

Predictive analytics can help predict the user load by analyzing past data. Organizations can use this data to better prepare to handle the predicted user load and provide experience assurance, which would help reduce customer churn. This data can help organizations better plan their future IT infrastructure requirements and capacity utilization.

Predict application outages before they happen

Predicting application downtime or outages before they happen helps perform the needed maintenance on the application without any downtime. This can save an organization time, money, and much more. Before an application outage, the IT infrastructure leaves many indirect clues hours, or even days, before it dies. The predictive analytics model can learn those patterns and continue to monitor for similar occurrences, predicting future failures before they happen. With this predictive model in place, preventive action can be taken at the right time.

Applying predictive analytics to forecast application performance

Predictive analytics in application performance improvement focuses on three main areas – Forecasted User Load, Response Time prediction, and Infrastructure Assessment.

User Load Prediction

Traditionally organizations have relied on peak user traffic in the past to come up with the number of users that might access the application in the future. This model has its own limitations as it does not consider factors such as the emergence of new technologies, changes in user behavior, and other disruptive factors. By using AI/ML predictive analytics, businesses can avoid these pitfalls, as forecasting models can be built by analyzing user behavior in real-time.

Using the data provided by the APM we have built a model to predict the number of users who will access the application six months into the future. We used the hourly monitoring data from production monitoring for the past 2 years as input to the model. We found that the accuracy provided by the model was low when we used classification and regression models due to missing data and data imbalance.

The data was normalized to handle missing data and data wrangling performed to mitigate the data imbalance and fed to a neural network. This increased the accuracy of the model to approx. 75–80%. The model was then converted to a self-learning algorithm by retraining the model on live data which further increased the model accuracy by 8-10%.

Such accurate forecast helps decrease the customer churn rate.

Response Time Prediction

APM tools capture a lot of application data and hardware utilization details. This data can be leveraged to predict the application response time.

To predict the application response time for a given user load, Exploratory Data Analysis using the user load as the input provides low accuracy and correlation. Even adding the application server hardware utilization details to the inputs and using multilinear regression models does not help achieve the required accuracy.

However, when we add the database server hardware utilization details, network telemetry, and middleware utilization details to the inputs and use the ARIMA models, we find that the model accuracy may increase to more than 75%.

Infrastructure Capacity Prediction

Apart from building the models to predict the user load and response times, we have also developed a model that estimates whether a given piece of hardware can handle a given user load. By tweaking the input parameters to the model, we can also predict the infrastructure required to handle a particular user load. To achieve this, a model was built by leveraging the APM data from production for a period of two years.

With a 75% accurate model, we then modified the hardware configuration to predict the application’s health.

Data is the new oil – are you analyzing it?

According to Forrester “They say data is the new oil. They say data is the new currency. They say data is the key competitive differentiator. All true. But the reality is sobering: Only 7% of firms report advanced, insights-driven practices.”

Businesses need to spend some time to realize the benefits. The predictive analytics implementation needs time as data has to be prepared, cleansed, and correlated, the right algorithm must be chosen to process the data, the algorithmic output must be understood, and the algorithm must be retrained with new sets of data and optimized until the required success factors are met. We can accelerate the deployment time by leveraging our experience with a similar application or domain.

Since predictive analytics depend on the data to predict future outcomes, any problems with the data quantity or quality will adversely impact the results. We can mitigate this problem by integrating our models with the APM tool in production.

To simplify the implementation of predictive analytics, we at Cigniti are developing a process that will monitor key performance indicators and develop its own unique system for setting dynamic thresholds. Essentially, the machine-based learning process will figure out high and low thresholds for when it’s appropriate to alert the operations team. The new model would shift these thresholds over time as it learns from aggregating historical data. It will allow for thresholds based on seasonality and history rather than a single static threshold set by a human.

Conclusion

Predictive analytics can predict future behavior and provide certain recommendations, but it cannot provide a complete course of action to handle the various scenarios. Organizations will have to plan how to handle the various outcomes coming out of the Predictive analytics models.

Connect with our experts today to resolve your Performance Engineering and Management needs.