Market Radar: Cloud-native Application Performance Management

Market Radar: Cloud-native Application Performance Management ? To manage the sheer scale of components in modern distributed systems, automation is becoming a necessity, and this opens the possibility for machine learning to power this automation. ? APM solutions will need to manage a range of types of applications, from monolithic to microservices, and various shades of hybrid in between. Recommendations Recommendations for enterprises All organizations should consider initiating proof of concepts for cloud-native development if they have not already done so. This style of development is being used not only for new and innovative products and services, but also for modernizing legacy systems. However, cloud-native introduces new complexities and is not an easy style of development. This complexity requires APM to help run these systems. Users from every type of organization have a wide range of solutions to choose from in APM, from open source to premium solutions designed for running enterprise mission-critical systems in production. Recommendations for vendors All APM vendors must be aware of the changes in the field that the latest waves of technology are bringing. Modern APM solutions must scale well, and have licensing models that are compatible under these scales, working for both vendor and user. The field of distributed tracing and time series metrics analytics has grown quickly. The next wave of innovation we believe will be in intelligent automation. We hear from nearly all vendors that they are busy doing research in applying AI, with some already in the market with intelligent products. Our advice therefore is to examine how AI can make a difference to your product/service, because your rivals are already doing this. Defining and exploring cloud-native APM Definition and characteristics The new-generation software application architecture based on microservices is changing our perception of previous architectures. Within the microservices view of how objects are coupled, everything that came before microservices is essentially a monolith, even service-oriented architecture. Earlier Ovum reports define microservices and cloud-native development, our new term for the whole set of connected technologies, which we define as the combination of agile development, DevOps delivery, and microservices deployed in containers, all making use of cloud infrastructure. Cloud-native APM is APM targeting these cloud-native applications and systems. Business value and applications APM is a well-established and relatively mature IT discipline across legacy and monolithic application architecture. With microservices in production, APM has had to grow and extend to meet the © Ovum. All rights reserved. Unauthorized reproduction prohibited. Page 3 Market Radar: Cloud-native Application Performance Management challenges the new architecture introduces. The key difference is the dynamic nature of the new IT environment with highly distributed systems. In legacy environments, servers are expected to be long-lived and applications long-running, and only occasionally brought down for updates. This creates a static environment where the connections that exist between components are long-lived, and troubleshooting can rely on agents at fixed points in the environment. With microservices, the concept of an application is broken into communications and transactions that run between services, where the components (services) and infrastructure (containers, servers, and so on) may appear and disappear solely to support the transitory transactions. Servers are short-lived and are never changed in production. They always exist as source code (infrastructure as code), and changes are made by terminating them in production and rolling out new instances. In dynamic environments, an issue will therefore be linked to the instance of the environment at the time of the incident, and this continuous change poses a new challenge for APM. The other main challenge in cloud-native environments is the sheer scale of the number of components that need to be monitored. With an application broken into thousands of services, and traffic scaling achieved by horizontally cloning services, the output from metric generators can become a big data problem. The APM field has evolved to meet these challenges and this report is mainly concerned with these new aspects of the field. Figure 1 shows the range of APM coverage across traditional monolithic applications, to modern microservices, with the central field representing hybrid systems. Figure 1: Cloud-native APM technology stack Source: Ovum © Ovum. All rights reserved. Unauthorized reproduction prohibited. Page 4
Please complete the form to gain access to this content