Michael Daum

A new approach to modular database systems

In this paper we present our approach towards a modularized database management system (DBMS) whose components can be adapted at runtime and show the modularization of a DBMS beneath the record-oriented interface as a first step. Cross-cutting concerns like transactions pose thereby a challenge that we answer with aspect-oriented programming (AOP). Finally we show the implementation techniques that enable the exchange of database modules dynamically. Particularly with regard to stateful components we define a service adaptation process that preserves and transmits the components state.

Integrating CEP and BPM: how CEP realizes functional requirements of BPM applications (industry article)

Business Process Management Systems (BPMSs) found broad application in recent years and there have been first suggestions how Complex Event Processing (CEP) can support or complement BPMS functionality. Though some business suites already provide technical integration of BPMSs and CEP engines, many benefits still remain undiscovered. We describe classification of event-driven data processing having a BPMS as data source and we reveal some relevant challenges for CEP. This paper provides contributions on both technical and conceptual level. From a technical point of view, we discuss mission-critical performance issues that depend on the chosen integration approach of BPMSs, CEP engines, and database systems. Scalability of BPM applications can benefit from outsourcing BPM computations to in-DB calculations and in-CEP processing. From a functional/conceptual point of view, we emphasize current CEP techniques and how they realize functional requirements of concrete BPM use cases.

A mediated publish-subscribe system for inter-institutional process support in healthcare

Inadequate availability of patient information is a major cause for medical errors and affects costs in healthcare. Traditional information integration in healthcare does not solve the problem. For chronic diseases and multimorbidity, the significance of patient information availability is yet increasing. Applying a document-oriented paradigm to a mediated publish-subscribe infrastructure allows to foster inter-institutional information exchange in healthcare. The goal of the proposed architecture is to provide information exchange between strict autonomous healthcare institutions, bridging the gap between primary and secondary care, following traditional paper-based working practice. In a distributed healthcare scenario, the patient has to maintain sovereignty over any personal health information. Therefore, the proposed mediated publish-subscribe architecture essentially decouples the roles of information author and information publisher into distinct actors.

Stateful mobile modules for sensor networks

Most sensor network applications are dominated by the acquisition of sensor values. Due to energy limitations and high energy costs of communication, in-network processing has been proposed as a means to reduce data transfers. As application demands may change over time and nodes run low on energy, get overloaded, or simply face debasing communication capabilities, runtime adaptation is required. In either case, it is useful to be able to migrate computations between neighboring nodes without losing runtime state that might be costly or even impossible to recompute. We propose stateful mobile modules as a basic infrastructure building block to improve adaptiveness and robustness of in-network processing applications. Stateful mobile modules are binary modules linked on the node itself. Even more importantly, they can be transparently migrated from one node to another, thereby keeping statically as well as dynamically allocated memory. This is achieved by an optimized binary format, a memory-efficient linking process and an advanced programming support.

Semantics of a runtime adaptable transaction manager

Database Management Systems (DBMSs) that can be tailored to specific requirements offer the potential to improve reliability and maintainability and simultaneously the ability to reduce the footprint of the code base. If the requirements of an application change during runtime the DBMS should be adapted without a shutdown. Runtime-adaptation is a new and promising research direction to dynamically change the behavior of a DBMS. Especially the adaptation of the Transaction Manager (TM) states a challenge. In this paper, we present the session semantics of a runtime-adaptable TM. We define preliminaries and assumptions to activate the TM during sessions from a conceptual point of view. The advantages and disadvanteges of different approaches are discussed, especially regarding the occurence of ANSI SQL phenomena. From a technical point of view, we define requirements for the architecture of the TM and the DBMS that arose in our prototype.

Exploitation of event-semantics for distributed publish/subscribe systems in massively multiuser virtual environments

Triggered by the fast evolving technical capabilities for implementing distributed global scale applications, online games have grown to a huge industry in recent years. Particularly, Massive Multiuser Virtual Environments (MMVEs), which allow for simultaneous activity of thousands of players in a virtual world, have been tremendously successful. Current architectures, however, use centralized approaches, which obviously do not scale beyond a certain point. Distributed event-based systems are a promising approach to reach both, performing and scalable architectures. The potential of this approach can only be fully exploited if event semantics is used to optimize event handling. Existing approaches actually do this to some degree, but typically in a very application specific manner. There is no generally applicable framework for classifying events according to their relevant semantic properties. In this paper, we propose a generally applicable classification of events as a first step on the way to flexibly adaptable generic event management systems. We exemplify the relevance of our semantic properties by classifying typical events in an existing MMVE. We discuss existing optimization strategies based on our semantic classification and outline a corresponding architecture.

Efficient and cost-aware operator placement in heterogeneous stream-processing environments

Operator placement for distributed stream-processing systems is still a challenging problem that can be modeled as a Task Assignment Problem (TAP). Multiple objectives are relevant for the optimization in heterogeneous stream-processing systems as there are different capabilities of the underlying networks and stream-processing nodes. We present an approach based on linear programming relaxation and iterative deterministic rounding. It uses an efficient linearization approach for the quadratic objective function that results from the TAP.

Query Processing and System-Level Support for Runtime-Adaptive Sensor Networks

We present an integrated approach for supporting in-network sensor data processing in dynamic and heterogeneous sensor networks. The concept relies on data stream processing techniques that define and optimize the distribution of queries and their operators. We anticipate a high degree of dynamics and heterogeneity in the network, which is expected to be the case for wildlife monitoring applications. The distribution of operators to individual nodes demands several system level capabilities not available in current sensor node operating systems. In particular, we developed means for replacing software modules, i.e. small applications, on demand and without loss of status information. In order to facilitate this operation, we added a lightweight module support for the Nut/OS system and implemented a new memory management that uses tags for preserving state across module updates and node reboots.

Black-box determination of cost models' parameters for federated stream-processing systems

For distribution and deployment of queries in distributed stream-processing environments, it is vital to estimate the expected costs in advance. Having heterogeneous Stream-Processing Systems (SPSs) running on various hosts, the parameters of a cost model for an operator must be determined by measurements for each relevant combination of an SPS and hardware. This paper presents a black-box method that determines the parameters of appropriate cost models that regard system-specific behavior. For some SPSs, there might not be any appropriate cost model available due to the lack of internal knowledge. If no cost model is available for any reason, we provide and apply a non-parametric model.

An overlay network for integration of WSNs in federated stream-processing environments

This paper presents an approach for seamless integration of hosts in heterogeneous networks in the context of data-stream processing. The integration of multiple heterogeneous hosts from the sensor-node level up to the level of high-performance workstations is one of the most promising concepts for extensive and efficient analysis of streaming data. For controlling such a network, communication between hosts is needed, e.g., to initiate stream processing, to configure queries, and to transmit streams. One of the key challenges is global addressing and transparent yet efficient data exchange despite diverse, differently capable networks involved. For this purpose, we developed a cross-platform overlay network that enables transparent communication between autonomous stream-processing systems on different hosts in miscellaneous networks for both data streams and control commands. The system directly uses underlying native protocols within each network so that the most efficient communication method is applied. Furthermore, global addressing of instances of stream-processing systems and routing over the individual communication paths is provided.