Person:

Zegura, Ellen W.

Permanent Link

https://hdl.handle.net/1853/71879

Associated Organization(s)

Organizational Unit

School of Computer Science

Full item page

Publication Search Results

Now showing 1 - 10 of 15

On Scalable QoS Routing: Performance Evaluation of Topology Aggregation

(Georgia Institute of Technology, 1999) Hao, Fang ; Zegura, Ellen W.

A number of important questions remain concerning the scalability of networks with quality of service guarantees. We consider one of these questions: can QoS routing scale to large networks? To address this question, we evaluate performance of four QoS routing algorithms both with and without topology aggregation, based on simulations of relatively large, structured networks. Among our observations, we find --- contrary to intuition --- that topology aggregation does not always have a negative impact on routing performance. Aggregation can reduce the routing information fluctuation, increase stability, and thus benefit routing performance. We also propose two new methods of aggregating routing information. Our hybrid aggregation method has performance much better than conventional star aggregation and approaches unaggregated performance. Our weighted aggregation method, while intuitively appealing, offers mixed performance across topologies.
Optimal Allocation of Clients to Replicated Multicast Servers

(Georgia Institute of Technology, 1999) Ammar, Mostafa H. ; Zegura, Ellen W. ; Fei, Zongming

Server replication is an approach that is often used to improve the scalability of a service. One of the important factors in the efficient utilization of replicated servers is the ability to direct client requests to the `best" server, according to some optimality criteria. Recently, there have been several proposals for multicast services in which a server delivers information to multiple clients simultaneously. Such proposals include multicasting of web content, multicast-based video services (on-demand and pay-per-view style services), multicasting of database content and broadcast disks. The goal of many of these proposals is to use multicast to enhance the ability of the service to handle a large number of clients economically. Multicast servers may be replicated for several reasons: to distribute the load among on-demand multicast servers, to balance the `feedback" load on the servers or on entities along the multicast tree from the servers, or to select the server that is at the root of the `best" multicast routing tree. In this paper we first give a definition of the static multicast server selection problem, in which we assume a set of static clients and multicast servers and consider how one might produce an optimalallocation of the clients to the servers. We propose a transformation method for deriving multicast server selection algorithms from the traditional multicast routing algorithms. To investigate the dynamic behavior of client join and leave and the cost incurred during the process, we next define the dynamic multicast server selection problem, in which the potential clients join and leave the multicast session dynamically, and the goal is to produce an optimal allocation of clients to servers with an emphasis on how this allocation behaves over time. We formulate the problem as a Markovian Decision Process (MDP) and analyze the tradeoff between the cost of the multicast tree(s) and the transition cost of establishing and removing links from the tree(s). We also explore the effect of join/leave frequency on optimal policy. Our analysis leads to two heuristics which we use to propose a selection algorithm. The algorithm uses a very simple join and leave strategy yet still can generate low cost trees. Our simulation compares the performance of our proposed algorithm with various other multicast server selection algorithms.
Inter-Receiver Fair Multicast Communication over the Internet

(Georgia Institute of Technology, 1999) Zegura, Ellen W. ; Ammar, Mostafa H. ; Jiang, Tianji

Multicast protocols target applications involving a large number of receivers with heterogeneous data reception capabilities.To accommodate heterogeneity, the sender may transmit at multiple rates, requiring mechanisms to determine the rates and allocate receivers. In this paper, we develop a protocol to control the rate of a multicast session, with the goal of maximizing the inter-receiver fairness, an intra-session measure that captures the collective "satisfaction" of the session receivers. Our target environment is the Internet, where fair sharing of bandwidth must be achieved via end-system mechanisms and fairness to TCP is important. We develop and evaluate protocols to maximize this measure by maintaining a fixed-rate base group and a variable-rate group. We show that our schemes offer improvement over single-rate sessions, while maintaining TCP-friendliness.
Scalability Techniques in QoS Routing

(Georgia Institute of Technology, 1999) Hao, Fang ; Zegura, Ellen W.

One barrier to deployment of large-scale QoS networks is scalability of QoS routing. Two major sources of overhead incurred by QoS routing are frequent network state updates and QoS-based path computation. Researchers have proposed an impressive array of techniques to reduce these overheads. The goal of this paper is to survey scalability techniques, in the context of a taxonomy intended to illustrate the relationships between different approaches and highlight areas of open research. We also include selected performance results from our own work, to demonstrate the inherent tension between scalability and performance. Our results are obtained using simulations of networks with up to 200 nodes; such large topologies are critical for fully evaluating scalability techniques.
Hashing-Based Traffic Splitting Algorithms for Internet Load Balancing

(Georgia Institute of Technology, 1999) Cao, Zhiruo ; Wang, Zheng ; Zegura, Ellen W.

Replication of resources is a key technique for improving Internet performance. Effective use of replicated resources requires good load distribution schemes. We study the performance of several hashing schemes for distributing traffic while preserving the order of packets within a flow. Traffic distribution with per-flow ordering has applications that include balancing traffic across multiple Internet access links and balancing HTTP request load in a web server farm. While hashing schemes for load balancing have been proposed in the past, this is the first comprehensive study of performance using real traffic traces. We evaluate five direct hashing methods and one table-based hashing method. We find that hashing using a 16-bit CRC over the TCP five-tuple gives excellent load balancing performance. Further, load-adaptive table-based hashing using the exclusive OR of the source and destination IP addresses achieves comparable performance to the 16-bit CRC. Table-based hashing can also distribute load according to unequal weights. We also report on four other schemes with poor to moderate performance.
Multiple-Channel Multicast Scheduling for Scalable Bulk-data Transport

(Georgia Institute of Technology, 1998) Ammar, Mostafa H. ; Zegura, Ellen W. ; Donahoo, Michael J.

A key technique for allowing a server to handle a large volume of requests for file transfers is to multicast the data to the set of requesting clients. Typically, the paths from the server to the clients will be heterogeneous in bandwidth availability. Multiple-Channel Multicast (MCM) is an approach that can be used to handle this heterogeneity. In this approach the data is multicast over multiple channels, each addressed as a separate multicast group. Each receiver subscribes to a set of channels (i.e., joins the corresponding multicast groups) commensurate with its own rate capabilities. Of particular interest in the design of MCM schemes is the scheduling of data transmission across the multiple channels to accomodate asynchronous requests from clients. In this paper, we present and analyze a new multiple-channel multicast approach called Partition Organization (PO) scheduling. The scheme is designed to result in good reception efficiency when compared to existing proposals while improving on their performance when other measures of interest (which we introduce) are considered.
Network Support for Multicast Video Distribution

(Georgia Institute of Technology, 1998) Bhattacharjee, Samrat ; Calvert, Kenneth L. ; Zegura, Ellen W.

Multicast video distribution in a best-effort environment presents challenges to system designers, including heterogeneity in the bandwidth availability on the paths from the sender to the receivers and dynamic behavior in the network and set of receivers over time. Classic approaches to dealing with dynamic conditions involve adaptation at the sender (for unicast) and adaptation driven by the receivers (for multicast). Both approaches have limitations that affect the quality of video received. In this paper, we consider a third option for the location of adaptation, namely: in the network. We demonstrate that a modest amount of state and computation at network routers can yield significant performance gains for multicast video distribution. Our schemes maintain the advantages of receiver-based adaptation, while overcoming the limitation. Since the network applies the adaptation, the time and place for adaptation can better match network conditions. Further, the adaptation can occur more rapidly, without the need for route changes. Finally, the adaptation can occur at finer granularity, providing better quality and more graceful degradation to receivers.
A Novel Server Selection Technique for Improving the Response Time of a Replicated Service

(Georgia Institute of Technology, 1997) Bhattacharjee, Samrat ; Zegura, Ellen W. ; Ammar, Mostafa H. ; Fei, Zongming

Server replication is an approach often used to improve the ability of a service to handle a large number of clients. One of the important factors in the efficient utilization of replicated servers is the ability to direct client requests to the best server, according to some optimality criteria. In this paper we target an environment in which servers are distributed across the Internet, and clients identify servers using our application-layer anycasting service. Our goal is to allocate servers to clients in a way that minimizes a client's response time. To that end, we develop an approach for estimating the performance that a client would experience when accessing particular servers. Such information is maintained in a resolver that clients can query to obtain the identity of the server with the best response time. Our performance collection technique combines server push with client probes to estimate the expected response time. A set of experiments is used to demonstrate the properties of our performance determination approach and to show its advantages when used within the application-layer anycasting architecture.
An Architecture for Active Networking

(Georgia Institute of Technology, 1996) Bhattacharjee, Samrat ; Calvert, Kenneth L. ; Zegura, Ellen W.

Active networking offers a change in the usual network paradigm: from passive carrier of bits to a more general computation engine. The implementation of such a change is likely to enable radical new applications that cannot be foreseen today. Large-scale deployment, however, involves significant challenges in interoperability, security, and scalability. In this paper we define an active networking architecture in which user control the invocation of pre-defined, network-based functions through control information in packet headers. After defining our active networking architecture, we consider a problem (namely, network congestion) that may benefit in the near-term from active networking, and thus may help justify migration to this new paradigm. Given an architecture allowing applications to exercise some control over network processing, the bandwidth allocated to each application's packets can be reduced in a manner that is tailored to the application, rather than being applied generically. Our results show that the ability to gracefully adapt to congestion makes a good case for active networking.
Application-Layer Anycasting

(Georgia Institute of Technology, 1996) Ammar, Mostafa H. ; Zegura, Ellen W. ; Shah, Viren ; Fei, Zongming ; Bhattacharjee, Samrat

Server replication is a key approach for maintaining user-perceived quality of service within a geographically wide-spread network. The anycasting communication paradigm is designed to support server replication by allowing applications to easily select and communicate with the "best" server, according to some performance or policy criteria, in a group of content- equivalent servers. We examine the definition and support of the anycasting paradigm at the application layer, providing a service that maps anycast domain names into one or more IP addresses using anycast resolvers. In addition to being independent from network-layer support, our definition includes the notion of filters, functions that are applied to groups of addresses to affect the selection process. We consider both metric-based filters (e.g., server response time) and policy-based filters; we further allow filtering both at the anycast resolver and local to the anycast client. A key input to the filtering process is metric information describing the relative performance of replicated servers. We examine the use of various techniques for maintaining this information at anycast resolvers.