J. A. Rolia and K. C. Sevcik. The method of layers. IEEE Transactions on Software Engineering, 21(8):689--700, Aug. 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....which a whole system can be constructed. For instance, modern performance analysis advocates a separation of the load model and the machine model, an idea developed already in [69, 62, 63] and similar ideas are also applied in stochastic rendezvous networks [101] and layered queueing networks [87]. As another, specific example, suppose one wished to model a communication system where two partners communicate over some communication medium. The model should reflect this structure, i.e. it should consist of three interacting submodels, one for each partner, and one for the medium. The user ....

J.A. Rolia and K.C. Sevcik. The Method of Layers. IEEE Transactions on Software Engineering, 21(8):689--700, August 1995.

....for a SRN can be found with an iteration technique that computes a series of intermediate solutions using an MVA approximation and continues until the estimated throughputs converge. The method of layers has been proposed for solving complex systems with one or more layers of software servers [Rol92, Rol95]. Each layer is allowed to call services from the layer immediately below it, thus introducing simultaneous resource possession. To find a solution for the layered queuing network (LQN) model, an iterative algorithm is used for solving the layers until successive response time estimates converge. ....

Rolia, J.A., Sevcik, K.C., The Method of Layers, IEEE Transactions on Software Engineering, Vol. 21, No. 8, August 1995, 689-699.

....a useful step towards performance evaluation of distributed systems, such analysis does not readily apply to distributed object systems. The main drawback is that the server in client server systems is not allowed to make requests of its own to other servers. Researchers at University of Toronto [6] and Carleton University [4, 22] are developing layered queueing network (LQN) modeling techniques to overcome this limitation. In a parallel study, we have applied the LQN approach to the design reported in this paper [7] We are working on incorporating the LQN methods for performance and ....

J. Rolia and K. Sevcik. The Method of Layers. IEEE Transactions on Software Engineering, 21(8):689--700, August 1995.

....distributions suggesting long term dependencies. This has far reaching implications for the performance analysis of systems which are exposed to WAN traffic. Since these heavy tailed distributions often have infinite means, well known analytical tools based on Mean Value Analysis (MVA) [71, 106, 128] become meaningless. Recent work by Feldmann et al. 47] suggest that WAN 19 traffic can be robustly modeled by multifractal processes (see for example [44, 61] To our knowledge the equivalent of queueing theory for WANs does not exist, yet. Finding the appropriate mathematical tools is still ....

J. A. Rolia and K. C. Sevcik. The Method of Layers. IEEE Transactions on Software Engineering, 21(8):689--699, August 1995.

....corresponding task. Resources are represented by queue nodes in the model. One queue node imitates the waiting time of every task process, when there are no requests to the task. Such kind of the network queuing model is presented in [Hag91] The proposed method is close to the Method of Layers [Rol95]. However, it has some differences such as: construction and computation of the hierarchical task models differ from the hierarchical software models proposed in [Rol95] using TR model permits to construct, by a natural way, a set of separate, but interdependent performance models for DA ....

....task. Such kind of the network queuing model is presented in [Hag91] The proposed method is close to the Method of Layers [Rol95] However, it has some differences such as: construction and computation of the hierarchical task models differ from the hierarchical software models proposed in [Rol95]; using TR model permits to construct, by a natural way, a set of separate, but interdependent performance models for DA systems. A set of different mapping models can be constructed by considering tasks and resources at various decomposition levels; set of performance models can be derived ....

[Article contains additional citation context not shown here]

J.A. Rolia and K.C. Sevcik. The Method of Layers. IEEE Transactions on Software Engineering, vol. 21, No 8, pp. 689-700, August 1995.

....Predictive performance models for distributed application systems can be used by distributed application developers and performance management staff to make quantitative comparisons between software design and system configuration alternatives. Models and their performance evaluation techniques [9, 10, 11] can also be used in capacity planning to determine whether specific application objects should be placed in the same server or how server processes should be allocated to nodes for a given workload mix of application requests. The models require information about remote procedure call ....

J.A. Rolia and K.C. Sevcik. The method of layers. IEEE Transactions on Software Engineering, 21(8):689--700, August 1995.

....a useful step towards performance evaluation of distributed systems, such analysis does not readily apply to distributed object systems. The main drawback is that the server in client server systems is not allowed to make requests of its own to other servers. Researchers at University of Toronto [6] and Carleton University [4, 22] are developing layered queueing network (LQN) modeling techniques to overcome this limitation. In a parallel study, we have applied the LQN approach to the design reported in this paper [7] We are working on incorporating the LQN methods for performance and ....

J. Rolia and K. Sevcik. The Method of Layers. IEEE Transactions on Software Engineering, 21(8):689--700, August 1995.

....various enhancements have been introduced to model phenomena such as simultaneous resource possession, fork and join, blocking and synchronization. Hybrid modeling techniques allow to model contention both at hardware and software levels by combining approximate solutions and analytical methods [22]. However, the complexity of parallel systems limits the applicability of these techniques. Therefore, in spite of its computation and time requirements, simulation is extensively used as it imposes no constraints on modeling. Evaluating system performance via experimental measurements is the ....

J. Rolia and K.C. Sevcik. The method of layers. IEEE Transactions on Software Engineering, 21(8):689--700, August 1995.

No context found.

Jerome Rolia and Ken Sevcik, The Method of Layers, IEEE Transactions on Software Engineering, Vol. 21, No. 8, pp. 689-700, August 1995.

....Predictive performance models for distributed application systems can be used by distributed application developers and performance management staff to make quantitative comparisons between software design and system configuration alternatives. Models and their performance evaluation techniques [44, 45, 56] can also be used in capacity planning, to determine whether specific application objects should be placed in the same server, or how server processes should be allocated to nodes for a given workload mix of application requests. We build these models based on data collected about applications as ....

J.A. Rolia and K.C. Sevcik. The method of layers. IEEE Transactions on Software Engineering, 21(8):689--700, August 1995.

....and non asymptotic conditions, as well. 7] introduces Mean Value Analysis (MVA) for studying the queuing delays of application requests that compete for devices. MVA has been adapted to support Layered Queuing Models (LQMs) that take into account contention for software as well as hardware [8, 10]. These techniques are used to estimate the mean response times of methods and processes in distributed application systems. They take into account the nested demands and queuing delays that affect required threading levels. The outline of the paper is as follows. Section 2 describes the problem ....

....the problem of determining maximum process utilization as a NLP model. In the NLP model, the process utilization is the objective function and the device utilization limits act as constraints. A control algorithm is given in Section 3 that integrates the NLP with LQMs and the Method of Layers [8] to evaluate the process utilization and discover its maximum over all workload conditions. Section 4 shows how the results can be interpreted and used for planning replication and activation policy for distributed application systems. A summary and description of future work are given in Section ....

[Article contains additional citation context not shown here]

Rolia J., Sevcik K.: The Method of Layers. IEEE Transactions on Software Engineering, Vol. 21, No. 8 (1995), 689-700

....A server may become a software bottleneck, thus limiting the potential throughput of the system. This can occur even if the devices used by the process are not fully utilized. Workload characterization for distributed applications must capture these effects. A Layered Queueing Network (LQN) model [5, 11, 13] is an extended queueing network model that also specifies visits between processes so that their layered requests for service and hence layered contention effects are represented. The parameters of the layered queueing network model are those of the queueing network model: ffl customer classes, ....

J. A. Rolia and K. A. Sevcik. The method of layers. IEEE Transactions on Software Engineering, 21(8):689--700, August 1995.

No context found.

J. A. Rolia and K. C. Sevcik. The method of layers. IEEE Transactions on Software Engineering, 21(8):689--700, Aug. 1995.

No context found.

J. A. Rolia, K. C. Sevcik, The Method of Layers, IEEE Transactions on Software Engineering, 21, 8, 1995.

No context found.

. J. A. Rolia and K. C. Sevcik, The Method of Layers, IEEE Transactions on Software Engineering 21

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC