On Load Balancing in Multicomputer/Distributed Systems Equipped with Circuit or Cut-Through Switching Capability

نویسندگان

  • Ching-Chih Han
  • Kang G. Shin
  • Sang Kyun Yun
چکیده

For multicomputer or distributed systems that use circuit switching, wormhole routing, or virtual cut-through, 2 the communication overhead and the message delivery time depend largely upon link contention rather than upon the distance between the source and the destination. That is, a larger communication overhead or a longer delivery delay occurs to a message when it traverses a route with heavier traac than one with a longer distance and lesser traac. This characteristic greatly aaects the selection of routes for interprocessor communication and/or load balancing. We consider the load balancing problem in these types of systems. Our objective is to nd the maximum load imbalance that can be eliminated without violating the (traac) capacity constraint while keeping the maximum link contention as low as possible. We investigate the load balancing problem under various conditions. First, we consider the case in which the excess load on each overloaded node is divisible. We devise a network ow algorithm to solve this type of load balancing problem optimally in polynomial time. Next, we impose the realistic assumption that the system uses a speciic routing scheme so that the excess load transferred from an overloaded node to an underloaded node must use the route found by the routing scheme. For this case, we use a graph transformation technique to transform the system graph to another graph to which the same network ow algorithm can be applied to solve the load balancing problem optimally. Finally, we consider the case in which the excess load on each overloaded node is indivisible, i.e., excess load must be transferred as an entity. We show that the load balancing problem of this type becomes NP-complete, and propose a heuristic algorithm as a solution. Any opinions, ndings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reeect the views of the funding agencies. 2 The last two are collectively called the cut-through switching.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A semi distributed load balancing scheme for large multicomputer systems

In this paper, we propose a semi distributed approach, for load balancing in large parallel and distributed systems. The proposed scheme is a two level hierarchical scheme which partitions the interconnection structure of a multiprocessor system into independent symmetric regions. We consider interconnection structures belonging to the classical infinite families of distance transitive graphs. ...

متن کامل

Performance Evaluation of Static and Dynamic Load Balancing Schemes for a Parallel Computational Fluid Dynamics Software (CFD) Application (FLUENT) Distributed across Clusters of Heterogeneous Symmetric Multiprocessor Systems

Computational Fluid Dynamics (CFD) applications are “highly parallelizable” and can be distributed across a cluster of computers. However, because computation time can vary with the distributed part (mesh), the system loads are unpredictable and processors can have widely different computation speeds. Load balancing (and thus computational efficiency) across a heterogeneous cluster of processor...

متن کامل

Transient Minimization Within Static Var Compensated Distribution Systems

VAR support should be supplied as close to the load as possible to minimize transmission losses. For voltage control and for improvement of total power factor in a distribution system the circuit- breaker switched capacitor banks can be used. The problems with this solution are the voltage steps caused by the large sizes of the capacitor banks as well as the transients caused on insertion. Thyr...

متن کامل

On the Virtual Topology of Loosely Coupled Multi- Computers

The proposed research addresses the use of networked computers as a large-scale multicomputer, that is, a tool for parallel and distributed computing. Informally speaking, it deals with realizing the vision “The network is the computer” (McNealy, Sun Microsystems). We intend to address a broad definition, whereby a multicomputer is a generalization of a single computer, namely, a general purpos...

متن کامل

Semi-Distributed Load Balancing For Massively Parallel Multicomputer Systems

This paper presents a semi-distributed approach, for load balancing in large parallel and distributed systems, which is different from the conventional centralized and fully distributed approaches. The proposed strategy uses a tw<rlevel hierarchical control by partitioning the interconnection structure of a distributed or multiprocessor system into independent symmetric regions (spheres) center...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:
  • IEEE Trans. Computers

دوره 49  شماره 

صفحات  -

تاریخ انتشار 2000