Institute for Systems Research

Permanent URI for this communityhttp://hdl.handle.net/1903/4375

Browse

Author: search.filters.author.Roussopoulos, N.

Search Results

Now showing 1 - 10 of 25
  • Thumbnail Image
    Item
    Reduction of Materialized View Staleness Using Online Updates
    (1998) Labrinidis, Alexandros; Roussopoulos, N.; ISR; CSHCN
    Updating the materialized views stored in data warehouses usually implies making the warehouse unavailable to users. We propose a MAUVE, a new algorithm for online incremental view updates that uses timestamps and allows consistent read-only access to the warehouse while it is being updated. The algorithm propagates the updates to the views more often than the typical once a day in order to reduce view staleness. We have implemented MAUVE on top of the Informix Universal Server and used a synthetic workload generator to experiment with various update workloads and different view update frequencies. Our results show that all kinds of update streams benefit from more frequent view updates, instead of just once a day. However, there is a clear maximum for the view update frequency, for which view staleness is minimal.
  • Thumbnail Image
    Item
    An Architecture for a Mobile and Dynamically Extensible Distributed DBMS
    (1998) Rodriguez-Martinez, M.; Roussopoulos, N.; ISR; CSHCN
    This report proposes an architectural framework for the design and implementation of a Distributed Database Management System which integrates Java and the Java Runtime Environment with the static set of operations found in a traditional database execution engine. With such a system, we intend to study the benefits of mobility of DBMS code (functionality shipping). This functionality shipping occurs at several levels, namely the query (i.e. predicates), executor (i.e. join operator) and access methods (i.e. index scan) levels.
  • Thumbnail Image
    Item
    Adaptive Data Broadcast in Hybrid Networks
    (1997) Stathatos, K.; Roussopoulos, N.; Baras, John S.; ISR; CSHCN
    Hybrid networks combine multiple communication modes and are fast, emerging as the most viable solution for the ever increasing demand for bandwidth and data services. Taking advantage of this new technology, we are proposing a hybrid scheme which effectively combines broadcast for massive data dissemination and unicast for individual data delivery. The goal is to build highly scalable systems with small response time. In this paper, we describe a technique that continuously adapts the broadcast content to match the hot-spot of the workload. We show that the hot-spot can be accurately obtained by, monitoring the ``broadcast misses'' observed through direct requests.

    This is a major departure from all other broadcast optimization schemes which are handicapped by their total reliance on complete knowledge of both ``hits'' and ``misses''. We also show that the proposed adaptive scheme performs effectively even under very dynamic and rapidly changing workloads. Extensive simulation results demonstrate both the scalability and versatility of, the technique. Another basic result obtained in this paper is that the overall, system's throughput depends only on the size of the hot-spot and not on the volume of the workload. This has far reaching implications for very large scale and high volume wide area information systems.

  • Thumbnail Image
    Item
    The Placement of File Sub-system Data Streams
    (1996) Gupta, Sandeep; Baras, John S.; Roussopoulos, N.; ISR; CSHCN
    An abstraction called Cues was implemented [1] for providing data streams over the file sub-system in the Unix kernel. The data stream of a Cue can be used for transfer of data as well as for control of the data stream by composing such structures. The algorithm for scheduling the compound Cue sequences read and write operations for data and control Cues in the kernel and retains other properties of user process driven read-write. This is possible for the data stream as this abstraction is over the file sub-system and the data stream is run in the context of the user process that created the Cue. This report reconsiders the architectural placement of the Cues, based on the experience with design and text of implementations. The underlying theme is to try to keep the control and flexibility as it is with user processes and keep from tying it in the kernel or user space unless specifically required, in effect re-evaluating the placement of the implementation in the kernel.
  • Thumbnail Image
    Item
    Adaptive Data Broadcasting Using Air-Cache
    (1996) Stathatos, K.; Roussopoulos, N.; Baras, John S.; ISR; CSHCN
    In the Data AirWaves Project at University of Maryland, we are integrating Direct Broadcast Satellite (DBS) systems with terrestrial networks to provide a hybrid and effective communication substrate lying between data resources and remote/mobile user applications. Smooth integration of these two media balances the need for rapid data dissemination to very large numbers of clients and on-demand interactive data services. This paper describes the air-cache, a method for effective data broadcasting and an algorithm which rapidly adapts the content of the cache based on the ﲭisses which result in explicit (on- demand) data requests. Simulation results show that the hypothesis of adapting based only on the misses performs quite reasonably and has very little deviation from a system that has complete information - both hits and misses.
  • Thumbnail Image
    Item
    Integrated Network Management of Hybrid Networks
    (1996) Baras, John S.; Ball, Michael O.; Karne, Ramesh K.; Kelley, Stephen; Jang, Kap D.; Plaisant, C.; Roussopoulos, N.; Stathatos, K.; Vakhutinsky, A.; Valluri, J.; Whitefield, D.; ISR; CSHCN
    We describe our collaborative efforts towards the design and implementation of a next generation integrated network management system for hybrid networks (INMS/HN). We describe the overall software architecture of the system at its current stage of development. This network management system is specifically designed to address issues relevant for complex heterogeneous networks consisting of seamlessly interoperable terrestrial and satellite networks. Network management systems are a key element for interoperability in such networks. We describe the integration of configuration management and performance management. The next step in this integration is fault management. In particular we describe the object model, issues of the Graphical User Interface (GUI), browsing tools and performance data graphical widget displays, management information database (MIB) organization issues. Several components of the system are being commercialized by Hughes Network Systems.
      A revised version of this report has been published in
      Proceedings of the 1st Conference of Commercial Development of Space, Part One, pp. 345-350, Albuquerque, New Mexico, January 7-11, 1996.
  • Thumbnail Image
    Item
    Simple Calls for Flexible Constructs Using the Traditional File API
    (1996) Gupta, Sandeep; Baras, John S.; Kelley, Stephen; Roussopoulos, N.; ISR
    We present the design for a remote qos control interface to the transport protocol based on existing work for similar applications. This puts together the read/write calls from the traditional file system API and an additional primitive. The addition amounts to programming an operating system data- streaming service which may be provided as a system call or otherwise using the standard techniques. Put together these allow much more than the traditional call based control interface. The resulting interface simplifies the mechanisms for distributed control. Parts of this interface have also been implemented in our ongoing experiments with file transfer.
  • Thumbnail Image
    Item
    Cues: File Subsystem Data Streams
    (1996) Gupta, Sandeep; Baras, John S.; Kelley, Stephen; Roussopoulos, N.; ISR
    We present a system call which enable directing high performance data transfers with in-kernel streams. The streams are defined and run using abstract data types called Cues, formed over the Unix file subsystem. The system call, named cue() , returns a descriptor, which can be used to write requests to the newly created stream. These requests define the flow of the stream. The abstraction simplifies the design of applications that transfer large amounts of data from files or devices. It also enables high throughput when multiple transfers are in progress. Cue code is compact, modular, and portable. This model also results in a simple mechanism for remotely cueing data flow using standard connections with peer processes. The implementation and tests are also described in this paper.
  • Thumbnail Image
    Item
    Consistency and Performance of Concurrent Interactive Database Applications
    (1995) Stathatos, K.; Kelley, Stephen; Roussopoulos, N.; Baras, John S.; ISR; CSHCN
    In many modern database applications, there is an emerging need for interactive environments where users directly manipulate the contents of the database. Graphical user interfaces (GUIs) display images of the database which must reflect a consistent up-to-date state of the data with minimum perceivable delay to the user. Moreover, the possibility of several applications concurrently displaying different views of the same database increases the overall system complexity. In this paper, we show how design, performance and concurrency issues can be addressed by adapting existing database techniques. We propose the use of suitable display schemas whose instances compose active views of the database, an extended client caching scheme which is expected to yield significant performance benefits and a locking mechanism that maintains consistency between the GUIs and the database.
  • Thumbnail Image
    Item
    ICON: A System for Implementing Constraints in Object-based Networks
    (1994) Goli, Shravan K.; Haritsa, Jayant R.; Roussopoulos, N.; ISR; CSHCN
    In today's Network Management scenario, the network operator's interface to the network is through a Management Information Base (MIB). The MIB stores all management related data such as configuration information, performance statistics, and trouble logs and so on. Configuration management, which is at the core of network management, is implemented through the MIB in a three step process: making updates to MIB data elements, checking the validity of the updates, propagating the effects of the updates to the network elements. While all three steps need to be executed efficiently for the MIB to serve its intended goal, the second step of checking update validity is especially important from the management viewpoint. For example, if an operator mistakenly configures a ninth port on an eight port card, it is essential that the MIB should both detect and prevent this error. Allowing such operations to go through would have adverse impact on the performance of the network (since it increases the network management traffic). Therefore, we focus primarily on the problem of checking the validity of updates to MIB data elements, which can be viewed as a specific instance of the general problem of constraint management in database systems. We introduce the design of ICON (Implementing Constraints in Object-based Networks), a proposed constraint management system. In ICON, constrains are expressed through rules. Each rule is composed of an event, a condition, and an action. Occurrence of the event triggers the rule, the condition is a boolean check, and the action is executed if the condition is satisfied. Rules and events are also treated as objects so that they can be created, modified, and deleted like other objects, thus providing a uniform view of rules and events in an OO context. The OO paradigm results in an extensible and a reusable system. To our knowledge, not much work has been done in this area and this paper would trigger further research in this area.