F. Lange, R. Kroeger and M. Gergeleit, "JEWEL: Design and Implementation of a Distributed Measurement System", IEEE Transactions on Parallel and Distributed Systems, Nov. 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....used as input to the instrumentation process. A modified compiler uses the meta data generated by the structural analysis of the code to place software triggers. Other options for automatically adding sensors to software include: instrumented interfaces (e.g. RPC stubs) and instrumented libraries [18]. PARADE 13 PARADE is an environment that facilitates the generation of visualisations for concurrent programs. This environment, which is designed for use by programmers, consists of three major components. The first component of the environment is responsible for monitoring the program that is ....

Lange, F., Kroeger, R. & Gergeleit, JEWEL: Design and Implementation of a Distributed Measurement System. IEEE Transactions on Parallel and Distributed Systems, 1992. 3(6): p. 657-672.

....failure mode [SSK 01] that are insu#cient here. The measurements and measures required are time oriented measures of the entire group, not just of a particular node. The topic of distributed system measurement and performance evaluation has been considered in the literature [BJS 95, LKG92] However, the combination of fault injection coupled with distributed measurement to perform assessment of dependability characteristics in a distributed system implementation has not been previously addressed in the literature. To solve both problems, a novel global state based fault injection ....

F. Lange, R. Kroeger, and M. Gergeleit. JEWEL: Design and implementation of a distributed measurement system. IEEE Transactions on Parallel and Distributed Systems, 3:657--671, Nov 1992.

....a central station. During the past few years, there were several systems built and some proposed for measuring the performance and observing the behavior of distributed systems. Several experimental systems were designed for online monitoring of distributed systems. Of these INCAS TMP [7] JEWEL [8], Spearmints [9] and MCC ES Kit [5] are some of the notable hybrid (monitor consisting of hardware and software) monitoring systems. A proposal for an expert system based online performance management tool is presented in [10] The approach similar to the one presented in this paper is applied ....

F. Lange, R. Kroeger, and M. Gergeleit, "JEWEL: Design and implementation of a distributed measurement system," IEEE Transactions on Parallel and Distributed Systems, vol. 3, no. 6, pp. 657--671, November 1992.

....be injected. Second, due to the need to compute a global measure such as the blocking time for the entire group, a facility for distributed measurement and measure specification is needed. The topic of distributed system measurement and performance evaluation has been considered in the literature [3, 20]. However, we are not aware of any past work that has combined the use of fault injection with distributed measurements to assess dependability characteristics of a real system. To solve both problems, we use a novel global state based fault injection technique. The technique involves abstracting ....

F. Lange, R. Kroeger, and M. Gergeleit. JEWEL: Design and implementation of a distributed measurement system. IEEE Trans. on Parallel and Distrib. Systems, 3:657--671, Nov. 1992.

....can be used to evaluate the dependability of reliable distributed systems built from COTS components, as well as systems built with specialized hardware. 1.3. Existing Tools Many tools have been written for the purpose of observing and testing distributed systems. Some of them, such as JEWEL [4], provide only measurement capabilities. JEWEL provides a global framework for system observation and measurement based on user defined measures. Results of such observations can be analyzed and displayed in a real time fashion. Such tools can observe a system under study, but cannot control the ....

F. Lange, R. Kroeger, and M. Gergeleit, "JEWEL: Design and implementation of a distributed measurement system," IEEE Transactions on Parallel and Distributed Systems, vol. 3, pp. 657-671, November 1992.

....To derive QoS related parameters such as loss ratio, traffic information needs to be retrieved from relevant monitors. This is similar to performance monitoring in distributed systems and also has the so called synchronization problem in retrieving information from distributed measurement points [7, 11]. In a QoS monitoring system, traffic information must be retrieved in a synchronized manner, i.e. the information retrieved from different measurement points must be of the same part of the flow that has been selected for monitoring. To meet these challenges, a few approaches are proposed in ....

.... A common approach to the synchronization problem in distributed measurement systems is to synchronize all monitors to a certain reference time such as the coordinated universal time (UTC) This can be achieved by using either the Global Positioning System (GPS) 5] a global time base (GTB) [11], or the Internet network time protocol (NTP) However, additional hardware is required to receive time signal from the GPS or to implement the GTB. Furthermore, synchronization may not be accurate for QoS monitoring in a large scale network if NTP is used, due to various network delays. This ....

[Article contains additional citation context not shown here]

F. Lange, R. Kroeger, and M. Gergeleit, "JEWEL: Design and Implementation of a Distributed Measurement System", IEEE Trans. Parallel and Distributed Systems, vol. 3, no. 6, pp. 657-671, Nov. 1992.

....for the implementation of Web services are recently changing the perspective in the monitoring area, where significant research work has been accomplished in the last years. Several distributed instrumentation systems have achieved very interesting results, especially to limit monitoring intrusion [20, 21]. However, they require instrumenting either the source code or the binary of monitored applications, and tend to be language and platform specific. This is not suitable for the monitoring of either mobile agents or, more generally, Web based services that consist of distributed heterogeneous ....

F. Lange, R. Kroeger, and M. Gergeleit, "JEWEL: Design and Implementation of a Distributed Measurement System", IEEE Transactions on Parallel and Distributed Systems, Vol. 3, No. 6, Dec. 1992.

....There are three major types of performance measurement tools: profilers, visualizations, and search tools. Profile metrics[1, 6, 15, 22] associate a value with each component of a distributed or parallel application (frequently procedures) and are presented as sorted tables. Visualizations[8, 13, 14, 18, 23] explain application performance using pictures. Search tools[10, 17, 21] help users to manage performance data information overload by treating the problem of finding a performance bottleneck as a search problem. However, all of these tools focus on the measurement and analysis of a specific ....

F. Lange, R. Kroger, and M. Gergeleit, "JEWEL: Design and Implementation of a Distributed Measurement System," IEEE Transactions on Parallel and Distributed Systems, 3(6), 1992, pp. 657671.

....conditions are also related to our probing concept. The packet pair technique [28] is one such technique designed to estimate the bottleneck rate on a path. Remote measurement and monitoring of system performance has also been explored as part of the extensive work on distributed system monitoring [39, 30]. 8 Concluding Remarks As the Internet continues to grow, server replication will be increasingly important as a technique to scale services. The effective utilization of a set of replicated servers hinges upon the ability to appropriately allocate servers to clients. Simple techniques such as ....

F. Lange, R. Kroeger, and M. Gergeleit. Jewel: Design and implementation of a distributed measurement system. IEEE Transactions on Parallel and Distributed Systems, 3(6):657-- 671, Novemeber 1992.

....conditions are also related to our probing concept. The packet pair technique [20] is one such technique designed to estimate the bottleneck rate on a path. Remote measurement and monitoring of system performance has also been explored as part of the extensive work on distributed system monitoring [30, 21]. 7 Concluding Remarks As the Internet continues to grow, server replication will be increasingly important as a technique to scale services. The effective utilization of a set of replicated servers hinges upon the ability to appropriately allocate servers to clients. Simple techniques such as ....

F. Lange, R. Kroeger, and M. Gergeleit. Jewel: Design and implementation of a distributed measurement system. IEEE Transactions on Parallel and Distributed Systems, 3(6):657-- 671, Novemeber 1992.

....general purpose systems software. Many distributed ISs have been developed over the past decade (e.g. 3, 8, 5] usually as components of larger software toolkits for analysis of parallel distributed systems. Only a few have been ported to multiple platforms and made available to broader usage [13, 9, 7, 4, 6]. Most of these contain subsystems that whose services could be provided by a common distributed IS kernel. In this paper we present a basic, reference implementation of a portable and flexible IS called BRISK (Baseline Reduced Instrumentation System Kernel) that we have developed as a part of a ....

.... of flexibility, BRISK s design is based on a generic distributed IS model that defines three components of a distributed IS: 1) local instrumentation server (LIS) 2) IS manager (ISM) and (3) transfer protocol (TP) For performance reasons, we have based the BRISK LIS implementation on JEWEL s [6] internal and generic external sensors. Section 2 describes the objectives of BRISK and approaches taken in its design and implementation. Section 3 describes the architecture and implementation details of BRISK. Along with the description, we discuss approaches that provided the IS performance ....

F. Lange, R. Kroeger, and M. Gergeleit. JEWEL: Design and implementation of a distributed measurement system. IEEE Transactions on Parallel and Distributed Systems, 3(6):657-- 671, November 1992.

....efforts have lead to well established protocols and paradigms like RMON, SNMP, CMIP [25] 3. System monitoring in distributed systems. A spate of efforts in late 80 s and early 90 s led to development of tools like ARTS, Kato, Jade, Incas, Tmp, ZM4, PATOP, DETOP, TOPSYS, Maritxu [10] and Jewel [6]. The tools were customized to the target system, which were distributed, cluster, or high performance computers, but in almost all cases homogeneous. Recent projects underway like Netlogger [3] NWS [21] Remos [16] and Gloperf [4] have been built for metacomputing. 4. Integrated management ....

F. Lange, R. Kroeger, M. Gergeleit. JEWEL: Design and Implementation of a Distributed Measurement System. Technical Report, German National Research Center for Computer Science (GMD), 1992.

....by the off the shelf instrumentation system JEWEL is fed to a dynamic scheduler. The scheduler uses this information to adaptively control the real time system to be responsive to the variation of important system variables. Reference: http: borneo.gmd.de:80 RS Papers direct direct.html and [7,15] Dynamic Resource Scheduling RMON RMON monitors the resource usage for distributed multimedia systems running RT Mach. Information collected by the instrumentation system is used for adaptively managing the system resources through real time features of the operating system. Reference: ....

Lange, F., Reinhold Kroger, and Martin Gergeleit, "JEWEL: Design and Implementation of a Distributed Measurement System,". IEEE Transactions on Parallel and Distributed Systems, 3(6), November 1992, pp. 657-671. Also available on-line from http:// borneo.gmd.de:80/RS/Papers/JEWEL/JEWEL.html.

....improve the performance of their parallel programs. 1. Introduction The design of performance tuning tools for parallel programming has historically been focused on analysing the data collected in the execution of a parallel program. A number of tools, such as Paragraph [1] AIMS [2] and JEWEL [3], have been developed, which collect and report the performance data. They rely on the user to examine the collected data or visualised performance and compare them with expected values to identify performance problems. This type of tools face a challenge: identifying a performance bottleneck ....

F. Lange, R. Kroeger and M. Gergeleit, "JEWEL: Design and Implementation of a Distributed Measurement System", IEEE Transactions on Parallel and Distributed Systems, Vol. 3, No. 6, Nov. 1992, pp. 657--671.

....timing faults. With TPS, hard, essential and soft tasks (and, thus, the respective applications) can coexist in one system and can share resources. 3 Monitoring The scheduling decisions of the dynamic real time system are based on the information gathered by the online monitoring system JEWEL [7] that has been developed at GMD during the last years. JEWEL has been designed as a hybrid, event driven, on line monitoring and performance measurement tool. It provides support for multi level event processing and filtering as well as automatic instrumentation for object oriented programs. ....

....also support other interactive services like gaming, gambling, shopping etc. The control system must cope with the different load and real time characteristics of the different applications. Due to the scheduling components provided, and together with a sophisticated online monitoring system [7], the system is able to determine the required resources and to assign dynamically a sufficient and guaranteed amount of CPU time as well as network bandwidth to each application. Again, the scheduler has to deal with different types of tasks. It is a strong requirement from the end users, that a ....

F. Lange, R. Kr6ger, M. Gergeleit. "JEWEL: Design and Implementation of a Distributed Measurement System", 1EEE Transactions on Parallel and Distributed Systems, Dec. 1992

....these tools have reached their limits, as they don t address distribution and concurrency. In the following a number of event based software monitoring systems that explicitly address realtime issues will be summarizes. Event based Monitors In 1992 the distributed measurement system JEWEL [Lan92] was presented. JEWEL consists of a generic set of flexible components and is not limited to a specific application domain. The JEWEL components are a configurable graphical presentation system for online visualization of the behavior of the system under test, a central interactive experiment ....

....are discarded. The observed attribute may be changed during run time either by the observing event processing system or by the system under test itself. The first option requires that the event processing system has access to these attributes (as implemented with relevant table in the JEWEL system [Lan92]) while the second implies that the system under test is written to be aware of its own instrumentation. The observed attribute of procedures can be stored in a static data structure as the number and names of the procedures are known at compile (or link) time. Object Oriented Languages ....

Lange, F., R. Krger, M. Gergeleit. JEWEL: Design and Implementation of a Distributed Measurement System, IEEE Trans. on Parallel and Distributed Systems, Vol. 3, No. 6, pp. 657-671, Nov. 1992.

....property qualifies JewelDC for the observation of realtime activities, as the degree of determinism that can be achieved with a system like Windows NT is not affected by the overhead introduced by the additional sensors. Results from our experiments on other platforms (vxWorks and Mach 3. 0 [11], Windows CE in ongoing experiments) have shown that the overhead is in the same order of magnitude on these systems. 5 Conclusions Distributed object oriented frameworks simplify the development of distributed applications by providing a high abstraction level that hides (and automatically ....

F. Lange, R. Krger, M. Gergeleit: "JEWEL: Design and Implementation of a Distributed Measurement System", IEEE Trans. on Parallel and Distributed Systems, Vol. 3, No. 6, pp. 657-671, Nov. 1992.

....values but no information on individual events. Detailed traces of system related events with accurate timestamps are more appropriate to understand e.g. why the synchronization of two threads fails in certain cases. 2 Related Work Over the last decade event based performance monitoring tools [5, 6, 8, 9] have shown their value for performance evaluation of complex software systems. Typically, preparing a system for an in depth monitoring means inserting sensors according to a certain monitoring or measurement model. Such a model consists of a (simplified) system model and a specification which ....

....and a specification which parts and quantities within this model have to be observed. The ability to create such a model provides a maximum of flexibility to the user of the monitoring system. However, after building and deploying the first generation of such a flexible monitoring system, JEWEL [5], we learned an important lesson: a generic default measurement model is needed JEWEL and many other tools did not assume a fixed measurement model, instead they allowed to define an application specific measurement setup. But this definition takes a lot of time. To simplify the use of a ....

F. Lange, R. Krger, M. Gergeleit, JEWEL: Design and Implementation of a Distributed Measurement System, IEEE Trans. on Parallel and Distributed Systems, Vol. 3, No. 6, November 1992, pp. 657-671

....computing resources, timing faults should be detected early. 3.1 Event Processing As described in more detail in section 4, the application is instrumented, i.e. augmented with sensors that generate and signal events. Events are collected and filtered by an on line event processing component [LKG 92] This component can work synchronously or asynchronously to the application s execution. In synchronous mode it blocks the progress of the signaling application thread until event processing is done, whereas in the asynchronous mode a sensor just stores event information in a buffer that is ....

F. Lange, R. Kroeger, M. Gergeleit: "JEWEL: Design and Implementation of a Distributed Measurement System", IEEE Transactions on Parallel and Distributed Systems, Dec. 1992.

No context found.

F. Lange, R. Kroeger and M. Gergeleit, "JEWEL: Design and Implementation of a Distributed Measurement System", IEEE Transactions on Parallel and Distributed Systems, Nov. 1992.

No context found.

Lange, F., Kroeger, R. and Gergeleit, M. (1992) JEWEL: design and implementation of a distributed measurement system. IEEE Trans. Parallel Distrib. Syst., 3, 657--671.

No context found.

LANGE, F., KROEGER, R., et GERGELEIT, M. Jewel: design and implementation of a distributed measurement system . IEEE Transactions on Parallel and Distributed Systems, 3(6):657--671. 129

No context found.

F. Lange, R. Kroeger, and M. Gergeleit. JEWEL: Design and implementation of a distributed measurement system. IEEE Trans. on Parallel and Dist. Systems, 3(6):657--671, Nov. 1992.

No context found.

F. Lange, R. Kroeger, and M. Gergeleit. JEWEL: Design and implementation of a distributed measurement system. IEEE Trans. on Parallel and Dist. Systems, 3:657--671, November 1992.

No context found.

Lange F, Kroeger R, Gergeleit M. JEWEL: Design and implementation of a distributed measurement system. IEEE Trans. Parallel and Distributed Systems Nov. 1992; 3: No. 6, 657 -- 671.

First 50 documents

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