Systems following the Polyglot Persistence paradigm are on the rise and with them come new problems and challenges. A major challenge is the ability to automatically self-adapt to changing requirements and workloads. The most difficult and as yet rarely discussed form of adaptivity relates to cahanges to the underlying data composition and topology. The search for a topology suited best for a given set of requirements can be modelled as a complex optimization problem.
Systems following the Polyglot Persistence paradigm are on the rise and with them come new problems and challenges. A major challenge is the ability to automatically self-adapt to changing requirements and workloads. The most difficult and as yet rarely discussed form of adaptivity relates to cahanges to the underlying data composition and topology. The search for a topology suited best for a given set of requirements can be modelled as a complex optimization problem. This paper proposes and formalizes Blueprints, which we define as graphs representing functional units composed of (heterogeneous) data stores. Blueprints can be used as manageable, predefined building blocks to form the highly complex system topologies Poly- and MultiStores use internally. Subsequently, the optimization search space can be limited to a set of Blueprints, which can be matched against the Poly-/MultiStores’ requirements. Furthermore, we discuss System requirements and their impact on adaptivity decisions and identify future research directions building upon our Blueprint concept.