CAP-Oriented Design for Cloud-Native Applications

Abstract

Brewer's conjecture, and its resulting formalization as the CAP theorem, impose serious limitations on the consistency, availability and network partitioning tolerance characteristics of distributed systems. Despite its importance however, few works explicitly consider the implications of the CAP theorem in the design of applications, especially for applications that are designed natively for the Cloud. In order to address this need, in this work we propose a CAP-oriented design methodology for Cloud-native applications. For this purpose we build and extend our previous work on Cloud architectural patterns. Finally, we show how the methodology can be used in practice to design an application solution with desired CAP properties.

Systematic Literature Review of the Cloud-ready Software Architecture

“…Here are just few examples of our findings. Andrikopoulos et al (2012), Toffetti et al (2016), Wilder (2012), Kratzke and Peinl (2016), Kratzke and Quint (2017) are using "cloud-native" term in their works and each of them provides a definition for it. "Cloud-ready" is mentioned by Brown and Capern (2014) from IBM, Kavis (2014), andWeinman (2016).…”

“…"Cloud-aware" term is met in Open Datacenter Alliance report (Casper et al (2014)) and Sodhi et al (2011). Andrikopoulos et al (2012) define cloud-native applications as ". .…”

“…Definitions provided in Andrikopoulos et al (2012), Toffetti et al (2016) and (2012) are more universal and technology agnostic. Authors of (Leymann et al (2016)) argues, that term "cloud-native" should be applicable to applications developed using containers or microservices.…”

“…A lot of work in cloud-ready applications development area is done by team from Institute of Architecture of Application Systems (IAAS) of University of Stuttgart. Andrikopoulos et al (2012) evaluate the effect of design decisions on the consistency, availability and partitioning (CAP) properties of cloud-ready applications. Fehhling et al (2011Fehhling et al ( , 2014 works discuss various cloud computing application patterns, challenges of deploying these patterns in the cloud and proposes possible solutions to overcome these challenges.…”

“…For the past few years a cloud-native term is industrialized by companies like VMware (see Web, a) or Pivotal (see Web, b) and is used to define containerized applications or applications developed using microservices. This term was used in Andrikopoulos et al (2012) long time before microservices and Docker or similar solutions came up. Ambiguity of this term is also realized by Kratzke and Quint (2016), and Leymann et al (2016).…”