We measure the distribution of lifetimes for UNIX processes and propose a functional form that fits this distribution well. We use this functional form to derive a policy for preemptive migration, and then use a trace-driven simulator to compare our proposed policy with other preemptive migration policies, and with a non-preemptive load balancing strategy. We find that, contrary to previous reports, the performance benefits of preemptive migration are significantly greater than those of non-preemptive migration, even when the memorytransfer cost is high. Using a model of migration costs representative of current systems, we find that preemptive migration reduces the mean delay (queueing and migration) by 35 -- 50%, compared to non-preemptive migration. 1 Introduction Most systems that perform load balancing use remote execution (i.e. non-preemptive migration) based on a priori knowledge of process behavior, often in the form of a list of process names eligible for migration. Althoug...

this paper is a description of our implementation and our experiences using it

Suppose that n balls are placed into n bins, each ball being placed into a bin chosen independently and uniformly at random. Then, with high probability, the maximum load in any bin is approximately log n log log n . Suppose instead that each ball is placed sequentially into the least full of d bins chosen independently and uniformly at random. It has recently been shown that the maximum load is then only log log n log d +O(1) with high probability. Thus giving each ball two choices instead of just one leads to an exponential improvement in the maximum load. This result demonstrates the power of two choices, and it has several applications to load balancing in distributed systems. In this thesis, we expand upon this result by examining related models and by developing techniques for stu...

Abstract not found

Scheduling in the context of parallel systems is often thought of in terms of assigning tasks in a program to processors, so as to minimize the makespan. This formulation assumes that the processors are dedicated to the program in question. But when the parallel system is shared by a number of users, this is not necessarily the case. In the context of multiprogrammed parallel machines, scheduling refers to the execution of threads from competing programs. This is an operating system issue, involved with resource allocation, not a program development issue. Scheduling schemes for multiprogrammed parallel systems can be classified as one or two leveled. Single-level scheduling combines the allocation of processing power with the decision of which thread will use it. Two level scheduling decouples the two issues: first, processors are allocated to the job, and then the job's threads are scheduled using this pool of processors. The processors of a parallel system can be shared i...

This paper presents a method for qualitative and quantitative analysis of load sharing algorithms, using a number of well known examples as illustration. Algorithm design choices are considered with respect to the main activities of information dissemination and allocation decision making. We argue that nodes must be capable of making local decisions, and for this efficient state dissemination techniques are necessary. Activities related to remote execution should be bounded and restricted to a small proportion of the activity in the system. The quantitative analysis provides both performance and efficiency measures, including consideration of the load and delay characteristics of the environment. To assess stability, which is also a precondition for scalability,we introduce and measure load sharing hit-ratio, the ratio of remote execution requests concluded successfully. Using our analysis method, we are able to suggest improvements to some published algorithms.

Heterogeneous Process Migration is a technique whereby an active process is moved from one machine to another. It must then continue normal execution and communication. The source and destination processors can have a different architecture, that is, different instruction sets and data formats. Because of this heterogeneity, the entire process memory image must be translated during the migration. "Tui" is a migration system that is able to translate the memory image of a program (written in ANSI-C) between four common architectures (m68000, SPARC, i486 and PowerPC). This requires detailed knowledge of all data types and variables used with the program. This is not always possible in non-type-safe (but popular) languages such as ANSI-C, Pascal and Fortran. The important features of the Tui algorithm are discussed in great detail. This includes the method by which a program's entire set of data values can be located, and eventually reconstructed on the target processor. Perfo...

A process is an operating system abstraction representing an instance of a running computer program. Process migration is the act of transferring a process between two machines during its execution. Several implementations

This paper evaluates linear models for predicting the Digital Unix five-second host load average from 1 to 30 seconds into the future. A detailed statistical study of a large number of long, fine grain load traces from a variety of real machines leads to consideration of the Box-Jenkins models (AR, MA, ARMA, ARIMA), and the ARFIMA models (due to self-similarity.) We also consider a simple windowed-mean model. The computational requirements of these models span a wide range, making some more practical than others for incorporation into an online prediction system. We rigorously evaluate the predictive power of the models by running a large number of randomized testcases on the load traces and then data-mining their results. The main conclusions are that load is consistently predictable to a very useful degree, and that the simple, practical models such as AR are sufficient for host load prediction. We recommend AR(16) models or better for host load prediction. We implement an online host load prediction system around the AR(16) model and evaluate its overhead, finding that it uses miniscule amounts of CPU time and network bandwidth

It is well known that simple randomized load balancing schemes can balance load effectively while incurring only a small overhead, making such schemes appealing for practical systems. In this paper, we provide new analyses for several such dynamic randomized load balancing schemes. Our work extends a previous analysis of the supermarket model, a model that abstracts a simple, efficient load balancing scheme in the setting where jobs arrive at a large system of parallel processors. In this model, customers arrive at a system of n servers as a Poisson stream of rate #n, # < 1, with service requirements exponentially distributed with mean 1. Each customer chooses d servers independently and uniformly at random from the n servers, and is served according to the First In First Out (FIFO) protocol at the choice with the fewest customers. For the supermarket model, it has been shown that using d = 2 choices yields an exponential improvement in the expected time a customer spends in the syst...