Analysis and optimization of a queueing model of a real-time computer control system
Real time computer control system queueing model to optimize preemptible and nonpreemptible job priority assignment
Schrage, L.↗
SEARCH · Engineering Papers
Results for “job queue times”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
The effect of model uncertainty on some optimal routing problems
The effect of model uncertainties on optimal routing in a system of parallel queues is examined. The uncertainty arises in modeling the service time distribution for the customers (jobs, packets) to be served. For a Poisson arrival process and Bernoulli routing, the optimal mean system delay generally depends on the variance of this distribution. However, as the input traffic load approaches the system capacity the optimal routing assignment and corresponding mean system delay are shown to converge to a variance-invariant point. The implications of these results are examined in the context of gradient-based routing algorithms. An example of a model-independent algorithm using online gradient estimation is also included.
Mohanty, Bibhu↗
Conservative parallel simulation of priority class queueing networks
A conservative synchronization protocol is described for the parallel simulation of queueing networks having C job priority classes, where a job's class is fixed. This problem has long vexed designers of conservative synchronization protocols because of its seemingly poor ability to compute lookahead: the time of the next departure. For, a job in service having low priority can be preempted at any time by an arrival having higher priority and an arbitrarily small service time. The solution is to skew the event generation activity so that the events for higher priority jobs are generated farther ahead in simulated time than lower priority jobs. Thus, when a lower priority job enters service for the first time, all the higher priority jobs that may preempt it are already known and the job's departure time can be exactly predicted. Finally, the protocol was analyzed and it was demonstrated that good performance can be expected on the simulation of large queueing networks.
Nicol, David M.↗
Conservative parallel simulation of priority class queueing networks
A conservative synchronization protocol is described for the parallel simulation of queueing networks having C job priority classes, where a job's class is fixed. This problem has long vexed designers of conservative synchronization protocols because of its seemingly poor ability to compute lookahead: the time of the next departure. For, a job in service having low priority can be preempted at any time by an arrival having higher priority and an arbitrarily small service time. The solution is to skew the event generation activity so that the events for higher priority jobs are generated farther ahead in simulated time than lower priority jobs. Thus, when a lower priority job enters service for the first time, all the higher priority jobs that may preempt it are already known and the job's departure time can be exactly predicted. Finally, the protocol was analyzed and it was demonstrated that good performance can be expected on the simulation of large queueing networks.
Nicol, David↗
Performance Analysis of the NAS Y-MP Workload
This paper describes the performance characteristics of the computational workloads on the NAS Cray Y-MP machines, a Y-MP 832 and later a Y-MP 8128. Hardware measurements indicated that the Y-MP workload performance matured over time, ultimately sustaining an average throughput of 0.8 GFLOPS and a vector operation fraction of 87%. The measurements also revealed an operation rate exceeding 1 per clock period, a well-balanced architecture featuring a strong utilization of vector functional units, and an efficient memory organization. Introduction of the larger memory 8128 increased throughput by allowing a more efficient utilization of CPUs. Throughput also depended on the metering of the batch queues; low-idle Saturday workloads required a buffer of small jobs to prevent memory starvation of the CPU. UNICOS required about 7% of total CPU time to service the 832 workloads; this overhead decreased to 5% for the 8128 workloads. While most of the system time went to service I/O requests, efficient scheduling prevented excessive idle due to I/O wait. System measurements disclosed no obvious bottlenecks in the response of the machine and UNICOS to the workloads. In most cases, Cray-provided software tools were- quite sufficient for measuring the performance of both the machine and operating, system.
Bergeron, Robert J.↗
Markov reward processes
Numerous applications in the area of computer system analysis can be effectively studied with Markov reward models. These models describe the behavior of the system with a continuous-time Markov chain, where a reward rate is associated with each state. In a reliability/availability model, upstates may have reward rate 1 and down states may have reward rate zero associated with them. In a queueing model, the number of jobs of certain type in a given state may be the reward rate attached to that state. In a combined model of performance and reliability, the reward rate of a state may be the computational capacity, or a related performance measure. Expected steady-state reward rate and expected instantaneous reward rate are clearly useful measures of the Markov reward model. More generally, the distribution of accumulated reward or time-averaged reward over a finite time interval may be determined from the solution of the Markov reward model. This information is of great practical significance in situations where the workload can be well characterized (deterministically, or by continuous functions e.g., distributions). The design process in the development of a computer system is an expensive and long term endeavor. For aerospace applications the reliability of the computer system is essential, as is the ability to complete critical workloads in a well defined real time interval. Consequently, effective modeling of such systems must take into account both performance and reliability. This fact motivates our use of Markov reward models to aid in the development and evaluation of fault tolerant computer systems.
Smith, R. M.↗
Dynamic resource allocation scheme for distributed heterogeneous computer systems
This invention relates to a resource allocation in computer systems, and more particularly, to a method and associated apparatus for shortening response time and improving efficiency of a heterogeneous distributed networked computer system by reallocating the jobs queued up for busy nodes to idle, or less-busy nodes. In accordance with the algorithm (SIDA for short), the load-sharing is initiated by the server device in a manner such that extra overhead in not imposed on the system during heavily-loaded conditions. The algorithm employed in the present invention uses a dual-mode, server-initiated approach. Jobs are transferred from heavily burdened nodes (i.e., over a high threshold limit) to low burdened nodes at the initiation of the receiving node when: (1) a job finishes at a node which is burdened below a pre-established threshold level, or (2) a node is idle for a period of time as established by a wakeup timer at the node. The invention uses a combination of the local queue length and the local service rate ratio at each node as the workload indicator.
Liu, Howard T.↗
Network Queuing System, Version 2.0
Network Queuing System (NQS) computer program is versatile batch- and device-queuing facility for single UNIX computer or group of computers in network. User invokes NQS collection of user-space programs to move batch and device jobs freely among different computers in network. Provides facilities for remote queuing, request routing, remote status, queue-status controls, batch-request resource quota limits, and remote output return. Revision of NQS provides for creation, deletion, addition, and setting of complexes aiding in limiting number of requests handled at one time. Also has improved device-oriented queues along with some revision of displays. Written in C language.
Walter, Howard↗
Optimal routing and buffer allocation for a class of finite capacity queueing systems
The problem of routing jobs to K parallel queues with identical exponential servers and unequal finite buffer capacities is considered. Routing decisions are taken by a controller which has buffering space available to it and may delay routing of a customer to a queue. Using ideas from weak majorization, it is shown that the shorter nonfull queue delayed (SNQD) policy minimizes both the total number of customers in the system at any time and the number of customers that are rejected by that time. The SNQD policy always delays routing decisions as long as all servers are busy. Only when all the buffers at the controller are occupied is a customer routed to the queue with the shortest queue length that is not at capacity. Moreover, it is shown that, if a fixed number of buffers is to be distributed among the K queues, then the optimal allocation scheme is the one in which the difference between the maximum and minimum queue capacities is minimized, i.e., becomes either 0 or 1.
Towsley, Don↗
Autotasked Performance in the NAS Workload: A Statistical Analysis
A statistical analysis of the workload performance of a production quality FORTRAN code for five different Cray Y-MP hardware and system software configurations is performed. The analysis was based on an experimental procedure that was designed to minimize correlations between the number of requested CPUs and the time of day the runs were initiated. Observed autotasking over heads were significantly larger for the set of jobs that requested the maximum number of CPUs. Speedups for UNICOS 6 releases show consistent wall clock speedups in the workload of around 2. which is quite good. The observed speed ups were very similar for the set of jobs that requested 8 CPUs and the set that requested 4 CPUs. The original NAS algorithm for determining charges to the user discourages autotasking in the workload. A new charging algorithm to be applied to jobs run in the NQS multitasking queues also discourages NAS users from using auto tasking. The new algorithm favors jobs requesting 8 CPUs over those that request less, although the jobs requesting 8 CPUs experienced significantly higher over head and presumably degraded system throughput. A charging algorithm is presented that has the following desirable characteristics when applied to the data: higher overhead jobs requesting 8 CPUs are penalized when compared to moderate overhead jobs requesting 4 CPUs, thereby providing a charging incentive to NAS users to use autotasking in a manner that provides them with significantly improved turnaround while also maintaining system throughput.
Carter, R. L.↗
Scheduling revisited workstations in integrated-circuit fabrication
The cost of building new semiconductor wafer fabrication factories has grown rapidly, and a state-of-the-art fab may cost 250 million dollars or more. Obtaining an acceptable return on this investment requires high productivity from the fabrication facilities. This paper describes the Photo Dispatcher system which was developed to make machine-loading recommendations on a set of key fab machines. Dispatching policies that generally perform well in job shops (e.g., Shortest Remaining Processing Time) perform poorly for workstations such as photolithography which are visited several times by the same lot of silicon wafers. The Photo Dispatcher evaluates the history of workloads throughout the fab and identifies bottleneck areas. The scheduler then assigns priorities to lots depending on where they are headed after photolithography. These priorities are designed to avoid starving bottleneck workstations and to give preference to lots that are headed to areas where they can be processed with minimal waiting. Other factors considered by the scheduler to establish priorities are the nearness of a lot to the end of its process flow and the time that the lot has already been waiting in queue. Simulations that model the equipment and products in one of Texas Instrument's wafer fabs show the Photo Dispatcher can produce a 10 percent improvement in the time required to fabricate integrated circuits.
Kline, Paul J.↗