All experts blogs

The Elves of the Future: With Artificial Intelligence instead of Hats

Software and web applications are becoming increasingly complex. At the same time, end users have higher expectations of performance, functionality and availability. To meet these requirements, we use Application Performance Monitoring tools. These tools offer significant insight into the quality of solutions and support us in monitoring the systems – now with the help of artificial intelligence.

Searching for a Needle in a Haystack

A modern web application consists of a wide range of services that are each individually configured and operated in different environments. It is obvious that this kind of complex setting can lead to performance problems and software errors. This is why we proactively and reactively advise our clients on performance issues.

We act reactively when our clients are already facing performance problems and the customer experience on the website in question has considerably worsened as a result. This might mean longer loading times or even service interruptions. The root causes may be very different, such as a new release on the system or surrounding systems, incorrect configurations or problems with infrastructure. Whatever the reason for the interference, it is vital to identify it as quickly as possible. In the time before smart Application Performance Monitoring tools, we were forced to search through endless system logs – an endeavor almost as fruitless as looking for a needle in a haystack.

We support our customers proactively by carrying out load and performance tests. Load tests are performed on the system based on the expected number of users. The system is consequently exposed to a multiple of the expected traffic. We performed load tests in August this year for an e-commerce client who invests in a comprehensive marketing campaign every autumn and was eager to know whether the platform could withstand the sudden onslaught of interest. We usually perform load tests like these two months before peak demand is expected so that we can make any necessary adjustments to the code, infrastructure or surrounding systems in good time. 

Regardless of whether we are acting reactively or proactively, we use a wide range of tools for all performance analysis to get a full picture of the current situation. Performance problems and software errors must be identified early on and associated with the negative customer experience.

Keeping Track with Dynatrace

We have made great progress in analyzing performance problems and software errors using Dynatrace. Dynatrace is an all-in-one Application Performance Monitoring (APM) solution and the only solution on the market to combine software and infrastructure monitoring with the customer experience. In addition to monitoring the performance and functionalities of different server and infrastructure components, it analyzes every aspect and dependency related to the customer experience. If this is negative because a website user can only perform an action to a limited degree or cannot perform it at all, Dynatrace finds the root cause in seconds.

In the case of the e-commerce client for whom we performed initial load tests, we were able to use the monitoring results and system logs to establish that the platform would not have been able to cope with the expected load at the time of the initial tests. However, we were unable to determine exactly why the subsystem was failing under high load. It was only when the Dynatrace APM tool was in operation during the second automated load test that it quickly became clear that one of the interfaces to a third-party system was responding too slowly. The system was so overloaded by this that the servers could no longer react. Performing tests with Dynatrace gave us the opportunity to make adjustments to the code and infrastructure before going live.

What are the benefits of APM Dynatrace?

One of Dynatrace’s main features is that it can be highly automated. Dynatrace automates the instrumentation of the environment and the graphical representation of the required metrics. It recognizes all the services, processes and dependencies without requiring an administrator to intervene manually. Dynatrace uses artificial intelligence to determine what normal utilization is and how a platform normally functions so it can identify standard performance. In the event of any anomalies, Dynatrace can determine the correlations and implications for the end user and alert the people responsible. By continually deploying Dynatrace on our clients’ platforms, we are able to detect performance problems proactively as well as reactively. Dynatrace works much like an elf. It might not wear a hat but is just as invisible and has the added bonus of artificial intelligence.

More On This Topic In Our Dossier On Artificial Intelligence

Why Artificial Intelligence is Not a Quantum Leap but Merely Makes Our Data-driven Work More Effective

Recording, interpreting and acting on data. Anything that artificial intelligence could one day do for us, we must first understand ourselves and be able to do on a small scale. We need cross-functional teams who are entirely data-driven in their work and who can rely on persistent and customer data that is as standardized as possible. We have a long way to go to achieve this.

Sitecore Cortex: How Smart is it Really?

How smart are content management systems? And what do you need to do to reap the benefits from it? We looked behind the marketing facade and put Sitecore Cortex to the test.