A study on application level monitoring of container-based microservices

Abstract

Microservice application monitoring has its proven benefits i.e. detecting service failure early, discover system bottlenecks, managing complexity of the microservice application, optimisation of system resources at runtime. Some research has been done in monitoring microservice applications, however, a secondary study was needed to identify gaps in the industrial setting and use these results to propose a model for creation of a defacto standard.

In the course of this study, we created a carried out a systematic mapping study of application-level monitoring of containerised microservices to address the first objective object to analysing the current state of the art and trends in monitoring containerised microservice applications. Secondly, we used the results from the systematic mapping study to guide with the design of a survey to help us analyse the current gaps and challenges faced by industry practitioners when monitoring their containerised microservices, answering the second research question. Lastly, we used the results from the practitioner survey to propose a model for a defacto standard.

The results showed solution proposals, followed by validation research were the most used strategies by researchers. Evaluation research which consists of benchmarking and evaluation of existing proposals and solutions needed more attention. More collaboration was needed between researchers and practitioners to foster industry adoption. We discovered that there was more need to address monitoring security aspects of a containerised microservice applications, monitoring across distributed systems, provision of microservice application migration and the monitoring of this migration process. During our survey analysis, we observed that companies that monitored their microservices experienced little to no downtime. However, they suggested the need for the ability for customisation of solutions, addressing of information gaps by enabling them create more infographics in addition to existing ones and the high cost of solutions that perform monitoring of applications at an application-level. From our insights gained while answering the two research questions answered thus far, there was need to address five elements that are key factors for driving a solution to becoming a de facto i.e. technology superiority, network effects, availability of complementary goods, pricing and installed base. Addressing these elements ensures organic growth by recommendations of early adopters to their peers, increased number of system users at any given time regardless of education background, skill level or computer literacy, avoidance of the need of clients to maintain multiple systems to address various aspects, friendly pricing of varying sizes of organisations and encouragement of strategic partnerships between solutions providers.