Simon Crosby

The Tempest-a practical framework for network programmability

The Tempest framework provides a programmable network environment by allowing the dynamic introduction and modification of network services at two levels of granularity. First, the switchlet and associated virtual network concepts enable the safe introduction of alternative control architectures into an operational network. The timescales over which such new control architectures can be introduced might vary from, for example, a video conferencing specific control architecture, which is active only for the duration of the conference, to a new version of a general purpose control architecture, which might be active for several months or longer. Second, the Tempest framework allows refinement of services at a finer level of granularity by means of the connection closure concept. In this case modification of services can be performed at an application-specific level. These attributes of the Tempest framework allows service providers to effectively become network operators for some well defined partition of the physical network. This enables them to take advantage of the knowledge they possess about how the network resources are to be used, by programming their own specially tailored control architecture. This, as our work with the Tempest shows, is a spur to creativity allowing many of the constraints imposed on operators and end users to be rethought and for new techniques to be quickly and safely introduced into working networks.

The Tempest: a framework for safe, resource assured, programmable networks

Most research in network programmability has stressed the flexibility engendered by increasing the ability of users to configure network elements for their own purposes, without addressing the larger issues of how such advanced control systems can coexist both with each other and with more conventional ones. The Tempest framework presented here extends beyond the provision of simple network programmability to address these larger issues. In particular, we show how network programmability can be achieved without jeopardizing the integrity of the network as a whole, how network programmability fits in with existing networks, and how programmability can be offered at different levels of granularity. Our approach is based on the Tempests ability to dynamically create virtual private networks over a switched transport architecture (e.g., an ATM network). Each VPN is assigned a set of network resources which can be controlled using either a well-known control system or a control system tailored to the specific needs of a distributed application. The first level of programmability in the Tempest is fairly coarse-grained: an entire virtual network can be programmed by a third party. At a finer level of granularity the Tempest allows user supplied code to be injected into parts of an operational virtual network, thus allowing application specific customization of network control. The article shows how the Tempest framework allows these new approaches to coexist with more conventional solutions.

Statistical properties of a near-optimal measurement-based CAC algorithm

Our algorithm, called Mosquito, allows sources to be ignorant of their statistics but offers near-optimal utilisation of the network. Our approach is based on large deviation theory: the large deviation rate-function (entropy) of bursty ATM traffic can be estimated from measurements of traffic activity. The entropy can be used to determine the bandwidth requirement of the traffic. In this paper, we explain the basic ideas behind the algorithm and describe its implementation. We present some results of a statistical investigation of the performance of the Mosquito CAC algorithm comparing it with that of various modifications of the algorithm.

Invited Practical connection admission control for ATM networks based on on-line measurements

We present algorithms for connections admission control (CAC) in an ATM network based upon the use of on-line measurement and estimation. (The algorithms described in this paper are the subject of a UK patent application.) Our approach uses the theory of large deviations, which is a valuable tool for reasoning about rare events in stochastic systems. We have shown previously that the large deviations rate function, or entropy, of ATM traffic can be estimated from activity measurements. This can be used to determine the effective bandwidth of the traffic. In this paper we present practical measurement-based CAC algorithms which are computationally efficient, and can be used with traffic whose statistical characteristics are unknown.

In-call renegotiation of traffic parameters

An approximate model of an asynchronous transfer mode (ATM) multiplexer is constructed and used to compare the performances of static and dynamic resource allocation schemes. Static allocation parameterizes a source at connection setup only, whereas dynamic allocation requires a source to inform the multiplexer whenever its bandwidth requirements change. A simple dynamic renegotiation protocol is analyzed, and bounds on the performance of an optimal dynamic protocol are derived. Markovian source models parameterized from working applications are used to contrast the behavior of the three control schemes.<<ETX>>

A comparative study of high speed networks

In the management of a modern LAN or campus network, two issues are of key importance, namely network performance and capacity planning. In this paper we report on results from an experimental programme which aims to quantify the performance that can be achieved with a real distributed application running over a range of different network technologies, including Ethernet, ATM, FDDI, and a 100 Mb/s packet-switched LAN. In particular we analyse the contribution of each of the various application and network components to the overall performance experienced by applications.

Measurement-based resource allocation for multimedia applications

Modern networks are now capable of guaranteeing a consistent quality of service (QoS) to multimedia traffic streams. A number of major operating system vendors are also working hard to extend these guarantees into the end-system. In both cases, however, there remains the problem of determining a service rate sufficient to ensure the desired quality of service. Source modeling is not a sustainable approach in the network case and it is even less feasible to model the demands of multimedia applications. The ESPRIT measure project is successfully using on-line measurement and estimation to perform resource allocation for bursty traffic in ATM networks. In this paper we consider the applicability of the same theory to resource allocation in a multimedia operating system which offers QoS guarantees to its applications.

The specification and verification of an experimental ATM signalling protocol

In this paper we present a formal specification of an experimental ATM signalling protocol using Temporal Logic of Transitions (TLT). The protocol allows users to request network services from an ATM-Master, which is responsible for control and management in a particular network domain. Robust and correct operation of the protocol is essential due to the use of unreliable RPC for communication and the instability of the underlying network. This was achieved by specifying the protocol compositionally, in terms of TLT programs with interfaces. This permitted each component to be verified individually using a temporal logic model checker, which releaved important errors in early versions of the specification. The paper includes a list of verified properties, eg. the absence of livelock and correct call termination, together with their verification times, to support this approach.

An experimental configuration for the evaluation of CAC algorithms

Interest in Connection Admission Control (CAC) algorithms stems from the need for a network user and a network provider to forge an agreement on the Quality of Service (QoS) for a new network connection. Traditional evaluation of CAC algorithms has been through simulation studies. We present an alternative approach: an evaluation environment for CAC algorithms that is based around an experimental test-rig. This paper presents the architecture of the test-rig and an evaluation of its performance.

Exploring the queueing behaviour of ATM switches

Abstract In this paper we describe methods which we have used to verify queueing models of two ATM switches, namely the Fairisle ATM switch developed at Cambridge University Computer Laboratory and an AT&T Globeview 2000 switch installed in the experimental test-bed at Telia Research Laboratories. Our approach relies on methods from large deviation theory, which we believe provides an appropriate scale on which agreement may be expected between a simple model and an actual ATM switch. We found that for the purpose of estimating delays the single-server queue provides an adequate model, at the large deviation scale, of both switches under the conditions investigated. However it does not reproduce the output behaviour of the Globeview switch at high loads. The model was evaluated using both analytic techniques and simulation. We emphasize that the model is not exact, indicating that very complex models may be required in order to obtain results more detailed than those afforded by large deviation theory. This calls into question the common practice of using detailed models of network traffic in conjunction with simplified models of the network itself.