Addressing BI Transactional Flows in the Real-Time Enterprise Using GoldenGate TDM

Abstract

It’s time to visit low latency and reliable real-time (RT) infrastructures to support next generation BI applications instead of continually debating the need and notion of real-time. The last few years have illuminated some key paradigms affecting data management. The arguments put forth to move away from traditional DBMS architectures have proven persuasive - and specialized architectural data stores are being adopted in the industry [1]. The change from traditional database pull methods towards intelligent routing/push models is underway, causing applications to be redesigned, redeployed, and re-architected. One direct result of this is that despite original warnings about replication [2] – enterprises continue to deploy multiple replicas to support both performance, and high availability of RT applications, with an added complexity around manageability of heterogeneous computing systems. The enterprise is overflowing with data streams that require instantaneous processing and integration, to deliver faster visibility and invoke conjoined actions for RT decision making, resulting in deployment of advanced BI applications as can be seen by stream processing over RT feeds from operational systems for CEP [3]. Given these various paradigms, a multitude of new challenges and requirements have emerged, thereby necessitating different approaches to management of RT applications for BI. The purpose of this paper is to offer a viewpoint on how RT affects critical operational applications, evolves the weight of non-critical applications, and pressurizes availability/data-movement requirements in the underlying infrastructure. I will discuss how the GoldenGate TDM platform is being deployed within the RTE to manage some of these challenges particularly around RT dissemination of transactional data to reduce latency in data integration flows, to enable real-time reporting/DW, and to increase availability of underlying operational systems. Real world case studies will be used to support the various discussion points. The paper is an argument to augment traditional DI flows with a real-time technology (referred to as transactional data management) to support operational BI requirements.