Logo des Repositoriums

Auflistung nach Autor:in "Petrov, Ilia"

1 - 6 von 6
  • Konferenzbeitrag
    Data-intensive systems on evolving memory hierarchies
    (INFORMATIK 2012, 2012) Petrov, Ilia; Bausch, Daniel; Gottstein, Robert; Buchmann, Alejandro
    New memory technologies with radically different properties are appearing. A substantial hardware and architectural redesign is required if they are to be used energy and performance efficient in a high-performance data-intensive system. In this paper we explore different alternatives and claim that the changing memory hierarchy and the challange for their energy efficient usage at the software layer disruptively impact the architecture and algorithms in any data-intensive system; direct control of the raw storage media is required yielding different hardware interfaces and software layers; data placement and versioning are emerging as new research fields promising most efficient usage in terms of performance and energy consumption.
  • Konferenzbeitrag
    Effective DBMS space management on native Flash
    (Datenbanksysteme für Business, Technologie und Web (BTW 2017), 2017) Hardock, Sergej; Petrov, Ilia; Gottstein, Robert; Buchmann, Alejandro
    In this paper we build on our research in data management on native Flash storage. In particular we demonstrate the advantages of intelligent data placement strategies. To e ectively manage physical Flash space and organize the data on it, we utilize novel storage structures such as regions and groups. These are coupled to common DBMS logical structures, thus require no extra overhead for the DBA. The experimental results indicate an improvement of transactional throughput for OLTP benchmarks of up to 60% and decrease in write-amplification of up to 2x, which doubles the longevity of Flash SSD. During the demonstration the audience can experience the advantages of the proposed approach on real Flash hardware.
  • Konferenzbeitrag
    Ein konzeptionelles Architekturrahmenswerk für Web-Anwendungen
    (Promise 2002 – Prozessorientierte Methoden und Werkzeuge für die Entwicklung von Informationssystemen, 2002) Jablonski, Stefan; Petrov, Ilia; Meiler, Christian
    Das Web gleicht einer sich beständig ändernden Technologielandschaft. Standards und Techniken, die zur Implementierung von Web-Anwendungen herangezogen werden, unterliegen ebenso wie die Plattformen, auf denen sie betrieben werden, einem kontinuierlichen Wandel. Um trotz dieser Dynamik strukturiert und systematisch Web-Anwendungen entwickeln zu können, fehlt eine saubere Entwurfsmethodik zur Entwicklung von Web-Anwendungen unter Berücksichtigung der zentralen Ziele Erweiterbarkeit und Flexibilität. Dieser Beitrag unterbreitet eine Begriffsdefinition für Web-Architekturen und ein konzeptionelles Architekturrahmenswerk für Web-Anwendungen. Mit diesem einhergehend werden wichtige Charakteristika eines solchen Rahmenswerks untersucht und ein Vorgehensmodell zur Anwendung vorgestellt.
  • Konferenzbeitrag
    Log-Data analysis through Tick-Discretisation
    (Digital Enterprise Computing (DEC 2016), 2016) Knödler, Christian; Moessner, Bernhard; Petrov, Ilia
  • Konferenzbeitrag
    MV-IDX: Multi-Version Index in Action
    (Datenbanksysteme für Business, Technologie und Web (BTW 2015), 2015) Gottstein, Robert; Goyal, Rohit; Petrov, Ilia; Hardock, Sergey; Buchmann, Alejandro
    Multi-Versioning DBMS (MV-DBMS) represent a very good match to the properties of Flash storage and the combination of both offers conceptual advantages. Yet, the specifics of indexing in MV-DBMS on Flash have been widely neglected. Although an index in a MV-DBMS references multiple versions of a data item, it is only allowed to return a single (at most one) version of that data item ”visible” to the current index operation. Logically separating version visibility checks from the index structure and operations, as in the traditional version-oblivious index, leads to version management overhead: to determine the appropriate version of a data item, the MV-DBMS first fetches all versions that match the search criteria and subsequently discards invisible versions according to the visibility criteria. This involves unnecessary I/Os to fetch tuple versions that do not need to be checked. We propose the idea that version-aware indexing has additional responsibility to recognize different tuple versions of a single data item and to filter invisible tuple versions in order to avoid unnecessary I/Os. In this work we demonstrate an approach called Multi-Version In- dex (MV-IDX) that allows index-only visibility checks which significantly reduce the amount of I/O as well as the index maintenance overhead. MV-IDX is implemented in the PostgreSQL open source MV-DBMS. We demonstrate that the MV-IDX achieves significantly lower response times and higher transactional throughput on OLTP workloads than the version-oblivious approach. We showcase latency and throughput improvements by utilizing the DBT2 TPC-C benchmarking tool and report saved I/Os. We also showcase how the proposed approach performs better on SSDs.
  • Konferenzbeitrag
    Near data processing within column-oriented dbmss for high performance analysis
    (Digital Enterprise Computing (DEC 2016), 2016) Vincon, Tobias; Petrov, Ilia