Benjamin Erb

Abstract: Distributed Reflective Denial of Service (DRDoS) attacks are an immanent threat to Internet services. The potential scale of such attacks became apparent in March 2018 when a memcached-based attack peaked at 1.7 Tbps. Novel services built upon UDP increase the need for automated mitigation mechanisms that react to attacks without prior knowledge of the actual application protocols used. With the flexibility that software-defined networks offer, we developed a new approach for defending against DRDoS attacks; it not only protects against arbitrary DRDoS attacks but is also transparent for the attack target and can be used without assistance of the target host operator. The approach provides a robust mitigation system which is protocol-agnostic and effective in the defense against DRDoS attacks.

Abstract: Stream-based graph systems continuously ingest graph-changing events via an established input stream, performing the required computation on the corresponding graph. While there are various benchmarking and evaluation approaches for traditional, batch-oriented graph processing systems, there are no common procedures for evaluating stream-based graph systems. We, therefore, present GraphTides, a generic framework which includes the definition of an appropriate system model, an exploration of the parameter space, suitable workloads, and computations required for evaluating such systems. Furthermore, we propose a methodology and provide an architecture for running experimental evaluations. With our framework, we hope to systematically support system development, performance measurements, engineering, and comparisons of stream-based graph systems.

Abstract: Distributed event-sourced systems adopt a fairly new architectural style for data-intensive applications that maintains the full history of the application state. However, the performance implications of such systems are not yet well explored, let alone how the performance of these systems can be improved. A central issue is the lack of systematic performance engineering approaches that take into account the specific characteristics of these systems. To address this problem, we suggest a methodology for performance engineering and performance analysis of distributed event-sourced systems based on specific measurements and subsequent, targeted optimizations. The methodology blends in well into existing software engineering processes and helps developers to identify bottlenecks and to resolve performance issues. Using our structured approach, we improved an existing event-sourced system prototype and increased its performance considerably.

Abstract: Several data-intensive applications take streams of events as a continuous input and internally map events onto a dynamic, graph-based data model which is then used for processing. The differences between event processing, graph computing, as well as batch processing and near-realtime processing yield a number of specific requirements for computing platforms that try to unify theses approaches. By combining an altered actor model, an event-sourced persistence layer, and a vertex-based, asynchronous programming model, we propose a distributed computing platform that supports event-driven, graph-based applications in a single platform. Our Chronograph platform concept enables online and offline computations on event-driven, history-aware graphs and supports different processing models on the evolving graph.

Abstract: Many data-driven applications require mechanisms for processing interconnected or graph-based data sets. Several platforms exist for offline processing of such data and fewer solutions address online computations on dynamic graphs. We combined a modified actor model, an event-sourced persistence layer, and a vertex-based, asynchronous programming model in order to unify event-driven and graph-based computations. Our distributed chronograph platform supports both near-realtime and batch computations on dynamic, event-driven graph topologies, and enables full history tracking of the evolving graphs over time.

Abstract: The increasing availability of 10G Ethernet network capabilities challenges existing transport layer protocols. As 10G connections gain momentum outside of backbone networks, the choice of appropriate TCP congestion control algorithms becomes even more relevant for networked applications running in environments such as data centers. Therefore, we provide an extensive overview of relevant TCP congestion control algorithms for high-speed environments leveraging 10G. We analyzed and evaluated six TCP variants using a physical network testbed, with a focus on the effects of propagation delay and significant drop rates. The results indicate that of the algorithms compared, BIC is most suitable when no legacy variant is present; CUBIC is suggested otherwise.

Abstract: In large-scale disaster scenarios, efficient triage management is a major challenge for emergency services. Rescue forces traditionally respond to such incidents with a paper-based triage system, but technical solutions can potentially achieve improved usability and data availability. We develop a triage management system based on commodity hardware and software components to verify this claim. We use a single-hop, ad-hoc network architecture with multi-master replication, a tablet-based device setup and a mobile application for emergency services. We study our system in cooperation with regional emergency services and describe experiences from a field exercise. We show that state-of-the-art commodity technology provides the means necessary to implement a triage management system compatible with existing emergency service procedures, while introducing additional benefits. This work highlights that powerful real-world ad-hoc networking applications do not require unreasonable development effort, as existing tools from distributed systems, like replicating No-SQL databases, can be used successfully.

Abstract: A basic reason for backend systems in mobile application architectures is the centralized management of state. Mobile clients synchronize local states with the backend in order to maintain an up-to-date view of the application state. As not all mobile social applications require strong consistency guarantees, we survey an alternative approach using special data structures for mobile applications. These data structures only provide eventual consistency, but allow for conflict-free replication between peers. Our analysis collects the requirements of social mobile applications for being suitable for this approach. Based on exemplary mobile social applications, we also point out the benefits of serverless architecture or architectures with a thin backend layer.

Abstract: While traditional graph computing usually employs batch processing, near-realtime computations on streaming data are often accomplished using event processing technologies. However, an increasing number of applications requires both computing capabilities for scenarios with highly dynamic and highly interconnected data fed by a stream of events. We suggest a novel platform architecture for graph computing that enables event-driven graph dynamicity while also supporting complex, long-running graph computations within the same system. To this end, we combine a vertex-based programming model, an asynchronous concurrency model using message passing, and an alternative persistence approach that is based on event sourcing. As a result, our approach allows for computations on dynamic live graphs as well as decoupled batch computations on graph snapshots. We also support the reconstruction of arbitrary previous graph states as well as temporal graph computations. Furthermore, our approach enables retroactive modifications of snapshot branches, so that alternate histories can be explored or future states of the graph can be simulated and predicted.

Tags: PROJEKT, AA, BA, MA, BenjaminErb, FrankKargl, sidgraph, distributed, DominikMeissner

Abstract: Blockchain technology allows for decentralized, distributed, and secure ledgers that store records (e.g., transactions). Popular blockchain-based systems such as Bitcoin and Etherum have emerged as so-called crypto-currencies. As the ledger maintains the full history of transactions, interactions within the system are always persisted. In this work, the student is asked to design and implement online and offline transaction analyses based on Chronograph, a data processing platform for evolving graphs developed at our Institute. Therefore, different blockchain-based systems should be surveyed and appropriate analysis mechanisms should be conducted.

Tags: PROJEKT, AA, BA, BenjaminErb, FrankKargl, sidgraph, distributed, DominikMeissner

Abstract: Event sourcing is an alternative persistence approach that maintains a log of state-changing events instead of altering states directly. Event-sourced architectures require an event store for efficiently appending and retrieving log entries. In this project, an evaluation of different key/value stores and alternative (No)SQL stores is to be conducted in order to identify stores appropriate for event sourcing.

Tags: PROJEKT, AA, BA, BenjaminErb, FrankKargl, sidgraph, distributed, DominikMeissner

Abstract: As part of an ongoing research project at our institute, we are currently developing a novel distributed computing platform prototype. The systems provides a graph-based, asynchronous programming model and takes advantage of event sourcing for history-aware computations. The aim of this student work is the identification and evaluation of appropriate secondary index structures, in order to provide fast access onto specific notes of the graph topology. Also, a prototypical implementation is part of this work.

Tags: AA, BA, BenjaminErb, FrankKargl, sidgraph, distributed

Abstract: As part of an ongoing research project at our institute, we are currently developing a novel distributed computing platform prototype. The systems provides a graph-based, asynchronous programming model and takes advantage of event sourcing for history-aware computations. The aim of this student work is to compile a survey of data mining mechanisms that can be executed on graphs. Specifically, asynchronicity has to be considered here. As a result, our current prototype should be evaluated on how it suits existing graph mining approaches.

Abstract: Distributed event-sourced systems adopt a fairly new architectural style for data-intensive applications that maintain the complete history of the application state. However, the performance implications of such systems is not yet well explored, let alone how the performance of these systems can be improved. A central issue is the lack of systematic performance engineering approaches that incorporate the specific properties of distributed event-sourced systems, such as messaging and event persistence. To address this problem, we developed a methodology for performance engineering and performance analysis of distributed event-sourced systems as part of a software engineering process. This approach helps developers to identify bottlenecks and resolve performance issues based on specific micro benchmarks and subsequent targeted optimizations. To show the practicality of our methology, we applied it to the \cg platform to improve the overall performance of its current research prototype. Using our structured approach, we improved the performance of the prototype system and made it more than twice as fast for certain workloads.

Abstract: Event sourcing is a style of software architecture wherein state altering operations to an application are captured as immutable events. Each event is appended to an event log, with the current state of a system derived from this series of events. This thesis addresses the utilization of retroactive capabilities in event-sourced systems: computing alternate application states, post hoc bug fixes, or the support of algorithms which have access to their own history, for example. The possibility of retroactively accessing and modifying this event log is a potential capability of an event-sourced system, but a detailed exploration how these operations can be facilitated and supported has not yet been conducted. We examine how retroaction can be applied to event-sourced systems and discuss conceptual considerations. Furthermore, we demonstrate how different architectures can be used to provide retroaction and describe the prototypical implementation of an appropriate programming model. These findings are applied in the Chronograph research project, in order to utilize potential temporal aspects of this platform.

Abstract: Man-made disasters, earthquakes, floods, and other natural disasters come with a great number of casualties, which have to be treated as quickly as possible by emergency services to minimize fatalities. Due to the large number of casualties and aid workers, it is difficult to maintain an accurate overview of the situation. To improve the clarity of the situation, a comprehensive support system can be used for the forces on-site, which supports them in information gathering and distribution to all involved parties. Previous work has failed to implement independency of public infrastructure (e.g. power grid, cellular network) or suffer data loss due to single node failures. To solve this problem, we propose a fault-tolerant design that fully distributes information to all devices in a mobile ad hoc network, while allowing offline work outside of it. We present a proof-of-concept prototype for the proposed design and show that its data distribution component behaves as designed using a series of trials. To the best of our knowledge, there is currently no DAN system that uses multi master replication to fully distribute data, where every node has an individual copy of every piece of information.

Abstract: This bachelor thesis presents a catalog of communication patterns for concurrent and distributed computations. We compose this catalog by reviewing the inter-process communication in common concurrency models and surveying existing pattern resources, such as professional books and weblogs. In addition to the selection and composition of patterns, we determine our own pattern template structure and an appropriate visualization, specifically matching the requirements of communication patterns. The catalog itself consists of a variety of patterns, intended for the reader to get a grasp of proven solutions for recurring problems in the field of concurrent programming. We provide simplified examples for every solution by the means of message-passing.

Abstract: NoSQL-Datenbanken werden immer häufiger produktiv eingesetzt, um die steigenden Datenmengen bewältigen zu können. Derzeit ist allerdings noch nicht klar, ob diese die notwendigen Sicherheitsanforderungen erfüllen. Zu diesem Zweck stellt die vorliegende Arbeit eine Methodik zur Sicherheitsanalyse von NoSQL-Datenbanken vor. Hierzu werden die wichtigsten Sicherheitsrisiken identifiziert und es wird aufgezeigt, wie eine NoSQL-Datenbank auf diese überprüft werden kann. Die Methodik wird auf die NoSQL-Vertreter Neo4j und CouchDB angewendet. Es zeigt sich hierbei, dass beide Datenbanken schwerwiegende Sicherheitsdefizite aufweisen. Deshalb werden Empfehlungen zur Steigerung der Informationssicherheit gegeben, die bei Beachtung die ermittelten Schwachstellen beheben.

Abstract: In dieser Arbeit wurde eine Plattform zur kollaborativen Dokumentation von Patterns entworfen und prototypisch implementiert. Hierfür wurden die Anforderungen für eine generische Plattform gesammelt und Konzepte für die kollaborative Nutzung erörtert. Die Implementierung verwendet verschiedene Konzepte des Web 2.0 zur Kollaboration und zeigt prototypisch den Funktionsumfang einer Plattform für Patterns verschiedener Anwendungsgebiete.

Abstract: In a common software system we occasionally want to know how we got to the current application state without making the system more complicated. Mostly, this question can not be answered because the software just saves the newest application state. Even if the software architects implement their own history support later, it can not tell anything about the previous changes. This is where Event Sourcing and Command Query Responsibility Segregation come into play. Event Sourcing saves every change as an event and Command Query Responsibility Segregation helps to handle the increased complexity. Therefore, we can build a system which provides a history support and which is still maintainable. If we use Event Sourcing and Command Query Responsibility Segregation, we just have to make small changes to a traditional architecture in order to fix that problem. By saving all changes as events we can evaluate everything we want.

Abstract: Datenbanksysteme stellen in Anwendungen die zentrale Komponente zur Persistierung von Daten dar. Die Speicherung der Daten kann über verschieden Datenmodelle, welche die Literatur in die Kategorien relational und NoSQL einteilt, realisiert werden. Die Datenmodelle bieten unterschiedliche Funktionalitäten in Bereichen wie Konsistenz, Verteilung und Skalierbarkeit. Skalierbarkeit stellt besonders für Datenbanken innerhalb Webanwendungen eine wichtige Anforderungen dar, da zum einen die Nutzeranzahl von Webanwendungen immer weiter steigt und zum anderen die Webanwendungen starke Lastschwankungen bewältigen müssen. Um diese Lastschwankungen verarbeiten zu können, werden flexible Ressourcen benötigt, die das Cloud-Computing verspricht. Diese Arbeit betrachtet die Cloud-Computing-Architektur des DBaaS, welche Datenbanken als abstrakte Ressource bereitstellt. Der Schwerpunkt liegt hierbei auf skalierenden DBaaS-Systemen. Das Ziel dieser Arbeit besteht aus dem Design und der Implementierung eines DBaaS-Dienstes, der eine automatisierte Skalierung bietet und auf frei verfügbarer Software basiert. Hierfür werden die Anforderungen eines solchen DBaaS-Dienstes anhand eines Anwendungsfalls herausgearbeitet und die Skalierbarkeit existierender Datenbanken auf Basis von Benchmarks untersucht. Aus diesen Ergebnissen wird ein prototypisches DBaaS-System umgesetzt.

Abstract: Gegenstand dieser Arbeit ist die Erstellung eines aktuellen Surveys von bestehenden Netzwerksimulatoren, insbesondere für VANET-Simulationen.

Abstract: Computational power for content filtering in high-speed networks reaches a limit, but many applications as intrusion detection systems rely on such processes. Especially signature based methods need extraction of header fields. Hence we created an parallel protocol-stack parser module on the NetFPGA 10G architecture with a framework for simple adaption to custom protocols. Our measurements prove that the appliance operates at 9.5 Gb/s with a delay in order of any active hop. The work provides the foundation to use for application specific projects in the NetFPGA context.

Tags: PROJEKT, AA, BA, MA, BenjaminErb, FrankKargl, sidgraph, distributed, DominikMeissner

Abstract: Blockchain technology allows for decentralized, distributed, and secure ledgers that store records (e.g., transactions). Popular blockchain-based systems such as Bitcoin and Etherum have emerged as so-called crypto-currencies. As the ledger maintains the full history of transactions, interactions within the system are always persisted. In this work, the student is asked to design and implement online and offline transaction analyses based on Chronograph, a data processing platform for evolving graphs developed at our Institute. Therefore, different blockchain-based systems should be surveyed and appropriate analysis mechanisms should be conducted.

Tags: PROJEKT, AA, BA, BenjaminErb, FrankKargl, sidgraph, distributed, DominikMeissner

Abstract: Event sourcing is an alternative persistence approach that maintains a log of state-changing events instead of altering states directly. Event-sourced architectures require an event store for efficiently appending and retrieving log entries. In this project, an evaluation of different key/value stores and alternative (No)SQL stores is to be conducted in order to identify stores appropriate for event sourcing.

Tags: PROJEKT, AA, BA, BenjaminErb, FrankKargl, sidgraph, distributed, DominikMeissner

Abstract: As part of an ongoing research project at our institute, we are currently developing a novel distributed computing platform prototype. The systems provides a graph-based, asynchronous programming model and takes advantage of event sourcing for history-aware computations. The aim of this student work is the identification and evaluation of appropriate secondary index structures, in order to provide fast access onto specific notes of the graph topology. Also, a prototypical implementation is part of this work.