Informatik
Refine
Document Type
- Conference proceeding (19)
- Journal article (7)
Language
- English (26)
Has full text
- yes (26)
Is part of the Bibliography
- yes (26)
Institute
- Informatik (26)
Publisher
- IARIA (26) (remove)
This paper reviews suggestions for changes to database technology coming from the work of many researchers, particularly those working with evolving big data. We discuss new approaches to remote data access and standards that better provide for durability and auditability in settings including business and scientific computing. We propose ways in which the language standards could evolve, with proof-of-concept implementations on Github.
Recent work on database application development platforms has sought to include a declarative formulation of a conceptual data model in the application code, using annotations or attributes. Some recent work has used metadata to include the details of such formulations in the physical database, and this approach brings significant advantages in that the model can be enforced across a range of applications for a single database. In previous work, we have discussed the advantages for enterprise integration of typed graph data models (TGM), which can play a similar role in graphical databases, leveraging the existing support for the unified modelling language UML. Ideally, the integration of systems designed with different models, for example, graphical and relational database, should also be supported. In this work, we implement this approach, using metadata in a relational database management system (DBMS).
At DBKDA 2019, we demonstrated that StrongDBMS with simple but rigorous optimistic algorithms, provides better performance in situations of high concurrency than major commercial database management systems (DBMS). The demonstration was convincing but the reasons for its success were not fully analysed. There is a brief account of the results below. In this short contribution, we wish to discuss the reasons for the results. The analysis leads to a strong criticism of all DBMS algorithms based on locking, and based on these results, it is not fanciful to suggest that it is time to re-engineer existing DBMS.
New storage technologies, such as Flash and Non- Volatile Memories, with fundamentally different properties are appearing. Leveraging their performance and endurance requires a redesign of existing architecture and algorithms in modern high performance databases. Multi-Version Concurrency Control (MVCC) approaches in database systems, maintain multiple timestamped versions of a tuple. Once a transaction reads a tuple the database system tracks and returns the respective version eliminating lock-requests. Hence under MVCC reads are never blocked, which leverages well the excellent read performance (high throughput, low latency) of new storage technologies. Upon tuple updates, however, established implementations of MVCC approaches (such as Snapshot Isolation) lead to multiple random writes – caused by (i) creation of the new and (ii) in-place invalidation of the old version – thus generating suboptimal access patterns for the new storage media. The combination of an append based storage manager operating with tuple granularity and snapshot isolation addresses asymmetry and in-place updates. In this paper, we highlight novel aspects of log-based storage, in multi-version database systems on new storage media. We claim that multi-versioning and append-based storage can be used to effectively address asymmetry and endurance. We identify multi-versioning as the approach to address dataplacement in complex memory hierarchies. We focus on: version handling, (physical) version placement, compression and collocation of tuple versions on Flash storage and in complex memory hierarchies. We identify possible read- and cacherelated optimizations.
The typed graph model
(2020)
In recent years, the Graph Model has become increasingly popular, especially in the application domain of social networks. The model has been semantically augmented with properties and labels attached to the graph elements. It is difficult to ensure data quality for the properties and the data structure because the model does not need a schema. In this paper, we propose a schema bound Typed Graph Model with properties and labels. These enhancements improve not only data quality but also the quality of graph analysis. The power of this model is provided by using hyper-nodes and hyper edges, which allows to present a data structure on different abstraction levels. We demonstrate by example the superiority of this model over the property graph data model of Hidders and other prevalent data models, namely the relational, object-oriented, and XML model.
In recent years, the Graph Model has become increasingly popular, especially in the application domain of social networks. The model has been semantically augmented with properties and labels attached to the graph elements. It is difficult to ensure data quality for the properties and the data structure because the model does not need a schema. In this paper, we propose a schema bound Typed Graph Model with properties and labels. These enhancements improve not only data quality but also the quality of graph analysis. The power of this model is provided by using hyper-nodes and hyper-edges, which allows to present data structures on different abstraction levels. We prove that the model is at least equivalent in expressive power to most popular data models. Therefore, it can be used as a supermodel for model management and data integration. We illustrate by example the superiority of this model over the property graph data model of Hidders and other prevalent data models, namely the relational, object-oriented, XML model, and RDF Schema.
Schema and data integration have been a challenge for more than 40 years. While data warehouse technologies are quite a success story, there is still a lack of information integration methods, especially if the data sources are based on different data models or do not have a schema. Enterprise Information Integration has to deal with heterogeneous data sources and requires up-to-date high-quality information to provide a reliable basis for analysis and decision-making. The paper proposes virtual integration using the Typed Graph Model to support schema mediation. The integration process first converts the structure of each source into a typed graph schema, which is then matched to the mediated schema. Mapping rules define transformations between the schemata to reconcile semantics. The mapping can be visually validated by experts. It provides indicators and rules to achieve a consistent schema mapping, which leads to high data integrity and quality.
This paper presents a concurrency control mechanism that does not follow a ‘one concurrency control mechanism fits all needs’ strategy. With the presented mechanism a transaction runs under several concurrency control mechanisms and the appropriate one is chosen based on the accessed data. For this purpose, the data is divided into four classes based on its access type and usage (semantics). Class O (the optimistic class) implements a first-committer-wins strategy, class R (the reconciliation class) implements a first-n-committers-win strategy, class P (the pessimistic class) implements a first reader-wins strategy, and class E (the escrow class) implements a firsnreaderswin strategy. Accordingly, the model is called OjRjPjE. Under this model the TPC-C benchmark outperforms other CC mechanisms like optimistic Snapshot Isolation.
This paper presents a concurrency control mechanism that does not follow a "one concurrency control mechanism fits all needs" strategy. With the presented mechanism a transaction runs under several concurrency control mechanisms and the appropriate one is chosen based on the accessed data. For this purpose, the data is divided into four classes based on its access type and usage (semantics). Class O (the optimistic class) implements a first-committer-wins strategy, class R (the reconciliation class) implements a first-n-committers-win strategy, class P (the pessimistic class) implements a first-reader-wins strategy, and class E (the escrow class) implements a first-n-readers-win strategy. Accordingly, the model is called OjRjPjE. The selected concurrency control mechanism may be automatically adapted at run-time according to the current load or a known usage profile. This run-time adaptation allows OjRjPjE to balance the commit rate and the response time even under changing conditions. OjRjPjE outperforms the Snapshot Isolation concurrency control in terms of response time by a factor of approximately 4.5 under heavy transactional load (4000 concurrent transactions). As consequence, the degree of concurrency is 3.2 times higher.
This work presents a disconnected transaction model able to cope with the increased complexity of longliving, hierarchically structured, and disconnected transactions. Wecombine an Open and Closed Nested Transaction Model with Optimistic Concurrency Control and interrelate flat transactions with the aforementioned complex nature. Despite temporary inconsistencies during a transaction’s execution our model ensures consistency.