M. Jones, P. Leach, R. Draves, and J. Barrera. Modular Real-Time Resource Management in the Rialto Operating System. In Proceedings of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....for individual requests. Priority driven services can generally be categorized this way, and are supported in real time kernels such as Al pha [8] and Mach [9] Under overload conditions, lower priority tasks are denied service in favor of more im portant tasks. In the Rialto operating system [10], a resource planner attempts to dynamically maximize user perceived utility of the entire system. However, the scheme does not adopt the notion of guaranteeing a reserved amount of resources for the application. Compromises between giving irrevocable service guarantees to arriving requests (in ....

M. B. Jones and P. J. Leach, "Modular real- time resource management in the rialto operating system," Technical Report MSR-TR-95-16, Microsoft Research, Advanced Technology Division, May 1995.

....provide any kind of CPU reservation; and QLinux [17] which provides hierarchical proportional share scheduling with bounded error. 6. RELATED WORK Previous middleware resource managers such as the QoS Broker [8] the Dynamic QoS Resource Manager [1] and the modular resource manager for Rialto [5] have tended to focus on adaptive applications and resource management in a distributed system. Unlike these systems, CRM is designed to be portable between OSs that provide di#erent scheduling abstractions, and takes the viewpoint that few desktop applications are capable of automatic ....

M. B. Jones, P. J. Leach, R. P. Draves, and J. S. Barrera, III. Modular Real-Time Resource Management in the Rialto Operating System. In Proc. of the 5th Workshop on Hot Topics in Operating Systems, May 1995.

....adapt to network and client variability via on line compression techniques. The technique, however, is inapplicable for dealing with server overload. 155 Novel real time operating systems and communication architectures were developed to embody QoS adaptation support. The Rialto operating system [64], targeted at multimedia applications, took the approach of dynamically maximizing aggregate system value using a resource planner. The Nemesis operating system designed in the context of the Pegasus project [78] investigated support for adaptive multimedia applications. In the multimedia ....

M. B. Jones and P. J. Leach, "Modular real-time resource management in the rialto operating system," Technical Report MSR-TR-95-16, Microsoft Research, Advanced Technology Division, May 1995.

....where tasks arriving later than others may be denied access to a resource even if they are more important [20] Ultimately, the admission control or resource management system should aim to provide maximum utility to the users of the system. To tackle these problems, the use of QoS managers [9, 12, 21] is usually proposed. Applications inform a QoS manager of their resource requirements and users may specify their respective utility values. The manager then attempts to maximise the overall resource utilisation and system utility. In [24] it is argued that this potentially NP hard problem can be ....

....an applications specific way. Applications may use utility functions to accomplish this task but it is not necessary to make utility functions explicit in the system. QoS managers are usually deployed to manage the distribution of resources amongst competing tasks. The resource planner in Rialto [9] and the Q RAM architecture [21] follow a centralised approach where tasks specify their resource requirements to a central entity responsible for resource management. Rialto deploys a simple negotiation protocol where tasks request a reservation and the planner either grants or rejects the ....

M. B. Jones, P. J. Leach, R. Draves, and J. S. Barrera. Modular Real-Time Resource Management in the Rialto Operating System. In Proc. of the 5th Workshop on Hot Topics in Operating Systems (HotOS-V), May 1995.

....MPEG decoder can reduce its requirements by scaling down the resolution of the stream. The renegotiation can also be used if a new stream is created by the application, which high priority requirements cannot be fulfilled with the current reservations. 5 Related Work The Rialto Operating System [7] defines a model for distributed real time resource management. A Resource 5 Provider exports two functions: it provides the resource (the data interface) and it provides an operation to determine the required lower level resources (the resource mapping part of the management interface) This ....

M. B. Jones, P. J. Leach, R. P. Draves, and J. S. B. III. Modular Real-Time Resource Management in the Rialto Operating System. In 5th Workshop on Tot Topics in Operating Systems, May 1995.

....of other activities on the system from a potentially malicious application component, and are desirable for the wider deployment of distributed component based applications. Existing approaches for enforcing qualitative and quantitative restrictions on resource usage rely on kernel support [JLDB95, MST94], binary modification [WLAG93] or active interception of the application s interactions with the operating system (OS) BG99, ET99, GWTB96] The kernel approaches are general purpose but require extensive modifications to OS structure, limiting their applicability for expressing flexible ....

....approaches can be classified into two broad categories: kernel level mechanisms and code transformation techniques. Kernel level mechanisms Real time Mach supports a Capacity Reserve abstraction [MST94] that guarantees applications a predictable CPU share over periodic time interval. Rialto [JLDB95, JR99] introduces CPU Reservation and Time Constraints, extending the NT kernel to support real time applications. Resource containers [BDM99] proposes a new UNIX kernel model for accounting and scheduling resources, which enables fine grained and predictable resource allocation. Eclipse [BGOS98] ....

M. Jones, P. Leach, R. Draves, and J. Barrera. Modular real-time resource management in the Rialto operating system. In Proc. of 5th Workshop on Hot Topics in Operating Systems, May 1995.

....protection [1, 3] These approaches have the limitation of lacking generality (because of reliance on specific programming languages and compilers) and are typically unable to enforce quantitative restrictions. Examples of the second class include approaches that rely on kernel support [13, 15], 1 binary modification [18] or active interception of the application s interactions with the operating system (OS) 2, 8, 9] for isolating resource usage. The kernel approaches are general purpose but require extensive modifications to OS structure and lack flexibility with respect to what ....

....some of these approaches lack generality because of reliance on specific programming languages and compilers. 2. 2 Enforcing compliance at run time Approaches for enforcing run time compliance of application behavior fall into two sub categories: Kernel level mechanisms such as CPU reservations [13, 15] and fair share queueing of CPU [10, 19] and network resources [6, 7, 20] have been employed, primarily in the context of real time operating systems, to enforce both qualitative and quantitative restrictions. Actually, such support provides a stronger guarantee of a certain level of resource ....

M. Jones, P. Leach, R. Draves, and J. Barrera. Modular real-time resource management in the Rialto operating system. In Proc. of 5th Workshop on Hot Topics in Operating Systems, May 1995. 19

....sites or client communities, resource containers must be combined with an appropriate scheduling policy for each type of resource (CPU, memory, disk and network bandwidth) Other related abstractions for resource principals have also been described in literature. Some examples are Activities [66] in the Rialto real time operating system, software performance units (SPU) 104] proposed in the context of shared memory multiprocessors, reservation domains in the Eclipse operating system [27, 26] and paths in the Scout [94] operating system. These are discussed in more detail in Section 5. ....

....performance isolation, it typically results in lower average utilization of cluster resources and higher average request latencies, because resources that are not currently utilized by one service class cannot be used by other service classes. Recent advances in operating systems research [66, 27, 104, 17, 26] allow e ective di erentiated QoS in single node Web servers. In this chapter, I address the problem of providing di erentiated QoS in cluster based network servers. I propose a new cluster wide abstraction called cluster reserve that is capable of achieving performance isolation between service ....

[Article contains additional citation context not shown here]

M. B. Jones, P. J. Leach, R. P. Draves, and J. S. Barrera. Modular real-time resource management in the Rialto operating system. In Proceedings of the Fifth Workshop on Hot Topics in Operating Systems (HotOS-V), Orcas Island, WA, May 1995. 122

....Approach Traditional systems, such as variants of Unix, attempt to solve only the global optimisation problem through policies embedded in the kernel. Recently in the literature the resource allocation problem has been approached by the introduction of a central Quality of Service Manager [10, 8, 4, 12]. This entity is made responsible for both the global and local optimisation of resource allocation, requiring accurate, detailed models of application behaviour. Various approximations to the solution are then obtained according to user policies. In general, this approach has several major ....

....the system currently has available, whilst still providing the user with the quality of service they require. 7 Related Work Resource management for Quality of Service has received a lot of attention in recent years. Examples include AQUA [8] the QoS Broker [11] the Resource Planner in Rialto [10] and QRAM [13] A good overview of work in this area is given in [2] Our work differs from the majority of these architectures in that we advocate a distributed solution to the problem, rather than 4 attempting to fully specify application requirements to a central QoS Manager. Various ....

R. Draves J. Barrera M. Jones, P. Leach. Modular real-time resource management in the Rialto operating system. In Proceedings of HotOS-V, May 1995.

....be preferable for the service provider to do the best it can (e.g. as defined by user priorities) than to fail at serving potentially important requests because of guaranteed resource commitment to current clients. The approach is suitable for soft real time systems. The Rialto operating system [22] adopts a similar philosophy. It attempts to dynamically maximize the users perceived utility of the entire system, rather than the performance of any particular application. Clients request their required resources from a resource planner whose goal is to compute a resource allocation that ....

M. B. Jones and P. J. Leach, "Modular real-time resource management in the rialto operating system," Technical Report MSR-TR-95-16, Microsoft Research, Advanced Technology Division, May 1995.

.... in resource delivery that was present with conventional operating system schedulers [Nieh93, Black94b] This has prompted researchers to design operating systems capable of supporting a mix of soft real time and conventional applications, for example SUMO Chorus [Coulson93] Rialto [Jones95] RT Mach [Mercer94] and the locally conceived Nemesis [Leslie96] These systems aim to allow applications to specify their Quality of Service (QoS) requirements to the operating system, which will then endeavour to meet them. Sometimes this will not be possible, but the system should then inform ....

....of the total resource. Users (or systems administrators) need a way of indicating how they wish resources to be shared out, but should not be presented with an incomprehensible array of knobs and dials that control the settings for each resource. Some form of QoS Agent [Nahrstedt95, Rajkumar98, Jones95, Leslie96] is required, perhaps akin to the X window system s window manager: Window managers share out screen real estate according to input from users, requests from applications, and policy set out in a (potentially auto generated) resource control le. In a similar fashion a QoS Agent could ....

M. Jones, P. Leach, R. Draves, and J. Barrera. Modular RealTime Resource Management in the Rialto Operating System. In Proceedings of HotOS-V, May 1995. (p 2)

....system [2] allows for service brokers , which translate applicationlevel resource requirements (quality metrics) to the system level. Rather than have explicit per application brokers, I propose to have the system learn to be a broker, by monitoring the application. The Rialto operating system [15] has, in its design, the notion of using user preferences to guide resource allocations between applications. Imprecise computations [6, 7, 14] support graceful degradation of real time systems under overload conditions. Each computation is modeled as a mandatory part followed by an optional ....

Michael B. Jones, Paul J. Leach, Richard P. Draves, and Joseph S. Barrera III. Modular real-time resource management in the rialto operating system. In Proceedings of the Fifth Workshop on Hot Topics in Operating Systems (HotOS-V), pages 12--17, Orcas Island, WA, USA, May 1995. IEEE Computer Society.

....the semantics of nice. One advantage of SPUs is that they were designed for use with shared memory multiprocessors. Extending the lottery scheduling framework for use with SMPs remains future work. Other systems have allowed applications to negotiate their resource usage with the operating system [Jon95, Nob97]. Our extended lottery scheduling framework lets applications coordinate their resource usage with each other, as well as with the system as a whole. Besides Waldspurger s own prototypes, others have implemented portions of the lottery scheduling framework [Arp97, Nie97] Petrou et al. Pet99] ....

Jones, M.B., Leach, P.J., Draves, R.P., Barrera, J.S., "Modular Real-Time Resource Management in the Rialto Operating System," Proc. of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.

....from taking resources from an important application or correcting suboptimal resource allocations; the QM remembers these corrections and attempts to track user preferences more closely in the future. Like the DQM, the QM focuses on adaptive applications. The modular resource manager for Rialto [36] divides the system into resource providers and activities. Each activity requests a resource set, a list of amounts of the different resources it requires to operate correctly, from a central planner. During overload, user preferences are consulted to resolve the conflict. The Rialto and HLS ....

Michael B. Jones, Paul J. Leach, Richard P. Draves, and Joseph S. Barrera, III. Modular RealTime Resource Management in the Rialto Operating System. In Proc. of the 5th Workshop on Hot Topics in Operating Systems, May 1995.

No context found.

M. Jones, P. Leach, R. Draves, and J. Barrera. Modular Real-Time Resource Management in the Rialto Operating System. In Proceedings of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.

No context found.

M. Jones, P. Leach, R. Draves, and J. Barrera. Modular Real-Time Resource Management in the Rialto Operating System. In Proceedings of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.

No context found.

M. B. Jones, P. J. Leach, R. P. Draves, and J. S. Barrera. Modular Real-Time Resource Management in the Rialto Operating System. In 5th Workshop on Hot Topic in Operating System (HotOS-V), May 1995.

No context found.

Jones, M.B., Leach, P.J., Draves, R.P., Barrera, J.S., "Modular Real-Time Resource Management in the Rialto Operating System," Proc. of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.

No context found.

Jones, M.B., Leach, P.J., Draves, R.P., Barrera, J.S., "Modular Real-Time Resource Management in the Rialto Operating System," Proc. of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.

No context found.

M. B. Jones, P. J. Leach, R. P. Draves, and J. S. Barrera. Modular real-time resource management in the Rialto operating system. In Proceedings of the Fifth Workshop on Hot Topics in Operating Systems (HotOS-V), Orcas Island, WA, May 1995.

No context found.

M. B. Jones, P. J. Leach, R. P. Draves, and J. S. Barrera. Modular real-time resource management in the Rialto operating system. In Proceedings of the Fifth Workshop on Hot Topics in Operating Systems (HotOS-V), Orcas Island, WA, May 1995.

No context found.

Michael B. Jones, Paul J. Leach, Richard P. Draves, and Joseph S. Barrera, III. Modular Real-Time Resource Management in the Rialto Operating System. In Proc. of the 5th Workshop on Hot Topics in Operating Systems, May 1995.

No context found.

Jones, M.B, Leach, P.J., Draves, R.P., et al. Modular real-time resource management in the Rialto operating system. In Proceedings of the 5th Workshop on Hot Topics in Operating Systems (HotOS-V), (Orcas Island, WA, May 1995).

No context found.

Jones, M.B., Leach, P.J., Draves, R.P., Barrera, J.S., "Modular Real-Time Resource Management in the Rialto Operating System," Proc. of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.

No context found.

Jones, M.B., Leach, P.J., Draves, R.P., Barrera, J.S., "Modular Real-Time Resource Management in the Rialto Operating System," Proc. of the Fifth Workshop on Hot Topics in Operating Systems, May 1995.

First 50 documents  Next 50

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