Theo Härder

Observations on optimistic concurrency control schemes

Abstract Optimistic concurrency control schemes allow uncontrolled access to shared data objects during transaction processing under the explicit assumption that read and write conflicts among transactions are rare events. Before a transaction commits, the DBMS has to validate that no conflict has occurred. Conflict resolution mainly relies on transaction abort. Two different optimistic concurrency control schemes are introduced and compared to each other. The problems of implementing such schemes and their implications on DBMS processing is investigated in some detail. A number of general properties of optimistic concurrency control schemes is derived, and their advantages and drawbacks w.r.t. two-phase locking approaches are discussed.

The personal electronic program guide—towards the pre-selection of individual TV programs

Digital and interactive TV systems will bring hundreds of TV program channels to the people. Then, the use of existing strategies for channel selection will fail: It is neither possible to select the program by “sur@g” through the channels, nor by using printed program guides. This paper describes a new approach towards creation of a Personal Electronic Program Guide (EPG) regarding the individual view preferences of a usez The foundation of individual TV program pre-selection is a classified and categorized TV program description and the maintenance of user profiles describing his viewing preferences. Besides creation of user projiles, an automatic update technique will be introduced. Channel selection events are taken as parameters for an adjustment of user projiles. Personal program pre-selection is done by using intelligent

Concurrency control issues in nested transactions

ltering techniques which apply individual user profiles to the TV program description (so-called matching). An approach to design and realize an EPG proto~pe system will be described in this papex

Node labeling schemes for dynamic XML documents reconsidered

The concept of nested transactions offers more decomposable execution units and finer-grained control over concurrency and recovery than “flat” transactions. Furthermore, it supports the decomposition of a “unit of work” into subtasks and their appropriate distribution in a computer system as a prerequisite of intratransaction parallelism. However, to exploit its full potential, suitable granules of concurrency control as well as access modes for shared data are necessary. In this article, we investigate various issues of concurrency control for nested transactions. First, the mechanisms for cooperation and communication within nested transactions should not impede parallel execution of transactions among parent and children or among siblings. Therefore, a model for nested transactions is proposed allowing for effective exploitation of intra-transaction parallelism. Starting with a set of basic locking rules, we introduce the concept of “downward inheritance of locks” to make data manipulated by a parent available to its children. To support supervised and restricted access, this concept is refined to “controlled downward inheritance.” The initial concurrency control scheme was based on S-X locks for “flat,” non-overlapping data objects. In order to adjust this scheme for practical applications, a set of concurrency control rules is derived for generalized lock modes described by a compatibility matrix. Also, these rules are combined with a hierarchical locking scheme to improve selective access to data granules of varying sizes. After having tied together both types of hierarchies (transaction and object), it can be shown how “controlled downward inheritance” for hierarchical objects is achieved in nested transactions. Finally, problems of deadlock detection and resolution in nested transactions are considered.

CFDC: a flash-aware replacement policy for database buffer management

We explore suitable node labeling schemes used in collaborative XML DBMSs (XDBMSs, for short) supporting typical XML document processing interfaces. Such schemes have to provide holistic support for essential XDBMS processing steps for declarative as well as navigational query processing and, with the same importance, lock management. In this paper, we evaluate existing range-based and prefix-based labeling schemes, before we propose our own scheme based on DeweyIDs. We experimentally explore its suitability as a general and immutable node labeling mechanism, stress its synergetic potential for query processing and locking, and show how it can be implemented efficiently. Various compression and optimization measures deliver surprising space reductions, frequently reduce the size of storage representation-compared to an already space-efficient encoding scheme-to less than 20-30% in the average and, thus, conclude their practical relevance.

An efficient infrastructure for native transactional XML processing

Flash disks are becoming an important alternative to conventional magnetic disks. Although accessed through the same interface by applications, flash disks have some distinguished characteristics that make it necessary to reconsider the design of the software to leverage their performance potential. This paper addresses this problem at the buffer management layer of database systems and proposes a flash-aware replacement policy that significantly improves and outperforms one of the previous proposals in this area.

Generic XMI-Based UML Model Transformations

Implementation techniques for relational database management systems (DBMSs) have proven their efficiency and robustness in many existing systems. However, many of these concepts and mechanisms cannot be used when implementing a native XML DBMS (XDBMS) because of substantial differences in the processing properties of natively stored XML documents as compared to relational tables. Therefore, we have to develop new and appropriate techniques with ACID transaction guarantees tailored to the processing characteristics of tree documents and the operations on them. For this reason, we want to provide for an efficient infrastructure of XDBMSs consisting of tree node addressing and indexing together with fine-grained locking of tree nodes. In this respect, our prime and novel contribution is to reveal the potential of our prefix-based node labeling called DeweyIDs supporting record addressing, indexing, and locking protocols. In this paper, we first sketch our version of prefix-based node labeling and summarize a quantitative study on them. An overview of our layered XDBMS architecture indicates the concepts and functionalities to be reused from relational DBMS implementations. The core part of the paper describes the infrastructural services for XML document storage with compressed DeweyIDs, the principles and methods for navigational and declarative processing of queries, as well as the lock modes and protocols to enable efficient collaboration. Selected empirical experiments evaluate the XTC system performance and support our system assessment.

The intrinsic problems of structural heterogeneity and an approach to their solution

XML-based Metadata Interchange (XMI) is an interchange format for metadata defined in terms of the MOF standard. In addition to supporting the exchange of complete models, XMI supports the exchange of models in differential form. Our paper builds on this feature to examine the possibility of XMI-based generic transformations of UML models. A generic transformation can be configured to generate (via XSLT) a specialized transformation that will be used to transform a UML model. The approach promotes model reuse, speeds up the modeling process and can be used to assure that only predefined semantics (as specialized by an agent) is included in the transformed model.

Generalizing prefix filtering to improve set similarity joins

Abstract. This paper focuses on the problems that arise when integrating data from heterogeneous sources in a single, unified database view. At first, we give a detailed analysis of the kinds of structural heterogeneity that occur when unified views are derived from different database systems. We present the results in a multiple tier architecture which distinguishes different levels of heterogeneity and relates them to their underlying causes as well as to the mapping conflicts resulting from the view derivation process. As the second essential contribution, the paper presents our approach to a mapping language solving the identified conflicts. The main characteristics of the language are its descriptiveness, its capability to map between schemas written in the relational, object-oriented, ER, or EXPRESS data model, and its facilities for specifying user-defined update operations on the view that are to be propagated to the data sources. Finally, we briefly discuss how this mapping information is employed to convert queries formulated with respect to the integrated view, into database operations over the heterogeneous data sources.

taDOM: A Tailored Synchronization Concept with Tunable Lock Granularity for the DOM API

Identification of all pairs of objects in a dataset whose similarity is not less than a specified threshold is of major importance for management, search, and analysis of data. Set similarity joins are commonly used to implement this operation; they scale to large datasets and are versatile to represent a variety of similarity notions. Most methods proposed so far present two main phases at a high level of abstraction: candidate generation producing a set of candidate pairs and verification applying the actual similarity measure to the candidates and returning the correct answer. Previous work has primarily focused on the reduction of candidates, where candidate generation presented the major effort to obtain better pruning results. Here, we propose an opposite approach. We drastically decrease the computational cost of candidate generation by dynamically reducing the number of indexed objects at the expense of increasing the workload of the verification phase. Our experimental findings show that this trade-off is advantageous: we consistently achieve substantial speed-ups as compared to known algorithms.