An adaptive algorithm for resolving processor thrashing in load distribution

Lu, Chin; Lau, Sau-Ming

doi:10.1002/cpe.4330070706

Cited by 12 publications

(6 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Processor thrashing refers to the situation when a receiver node is flooded with too many incoming tasks transferred from busy nodes, so that the receiver itself becomes overloaded inadvertently [8]. In [15], Shivaratri and Singhal used a simple method to prevent processor thrashing.…”

Section: Background Studymentioning

confidence: 98%

A Negotiation Protocol for Dynamic Load Distribution Using Batch Task Assignments

Lau

1998

Journal of Parallel and Distributed Computing

Self Cite

View full text Add to dashboard Cite

Section: Background Studymentioning

confidence: 98%

A Negotiation Protocol for Dynamic Load Distribution Using Batch Task Assignments

Lau

1998

Journal of Parallel and Distributed Computing

Self Cite

View full text Add to dashboard Cite

“…The majority of existing dynamic LD algorithms fall into this category [2,[5][6][7][8][9][10][11][12]. An important category of dynamic algorithms under Casavant and Kuhl's taxonomy is the dynamic suboptimal heuristic-based algorithms.…”

Section: Casavant and Kuhlmentioning

confidence: 99%

“…It means that a node attempting to identify a task transfer partner will send a query message to a selected target node to ask for its consent [2][3][4][5][6][7][8]. It means that a node attempting to identify a task transfer partner will send a query message to a selected target node to ask for its consent [2][3][4][5][6][7][8].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Dynamic load distribution using anti-tasks and load state vectors

Lau

Leung

1998

Concurrency: Pract. Exper.

Self Cite

View full text Add to dashboard Cite

Polling‐based load distribution (LD) algorithms suffer from two weaknesses: (i) load information exchanged during a polling session is confined to the two negotiating nodes only; (ii) as the distributed system grows in size (in terms of the number of constituent nodes), a larger number of polling sessions, and thus a higher amount of network bandwidth consumption and CPU overhead, are needed. We propose a new LD algorithm which is based on anti‐tasks and load state vectors. This new algorithm avoids the above weaknesses of polling‐based LD algorithms. Anti‐tasks are composite agents which travel around a distributed system to facilitate the pairing up of task senders and receivers, as well as the collection and dissemination of load information. Time‐stamped load information of processing nodes is stored in load state vectors which, when used together with anti‐tasks, encourage mutual sharing of load information among processing nodes. Anti‐tasks, which make use of load state vectors to decide their traveling paths, are spontaneously directed towards processing nodes having high transient workload, thus allowing their surplus workload to be relocated quickly. Using simulations, we evaluate the performance of our new algorithm by comparing its performance with a number of well‐known polling‐based load distribution algorithms. We found that our algorithm provides significant reduction of mean task response time over a large range of system sizes. The cost of achieving this performance gain in terms of CPU overhead and channel bandwidth consumption is generally comparable to the other algorithms we studied. © 1998 John Wiley & Sons, Ltd.

show abstract

“…The key in this protocol is the virtual workload model which ensures coordinated task dispersements within the distributed systems, while at the same time avoids processor thrashing [4].…”

mentioning

confidence: 99%

A negotiation protocol for batch task assignments in dynamic load distribution

Lau

1997

Proceedings of the 1997 ACM Symposium on Applied Computing - SAC '97

Self Cite

View full text Add to dashboard Cite

A dynamic load distribution (LD) algorithm improves the performance of a distributed system by using current system load information to relocate application tasks among processing nodes so that workload imbalance in the system can be smoothed out.In this paper, we propose a Guarantee and Reservation Protocol (GR Protocol) which can be used by dynamic LD algorithms. The protocol enables batch task assignments, which mean that multiple tasks can be selected for remote execution during each single sender-receiver negotiation session. This is in contrast to the commonly used single task assignments. Principally, the GR Protocol is used for obtaining the mutual agreement between a task sender and a receiver on the appropriate task batch size. The key in this protocol is the virtual workload model which ensures coordinated task dispersements within the distributed systems, and at the same time avoids processor thrashing. That is, to avoid situations when a receiver node is flooded with excessive incoming tasks and thus becomes overloaded.are needed before the workload imbalance can be smoothed out. There are two immediate performance penalties: (1) unnecessarily high CPU and communication overhead can be incurred; and (2) tasks cannot be dispersed fast enough, resulting in large task response times. A new LD algorithm has been proposed recently using batch task assignments, which mean that multiple tasks can be selected for remote execution during each single sender-receiver negotiation session [5]. This approach requires that the amount of workload contained in a task batch, referred to as the task batch size, must be determined dynamically during negotiations based on the load states of the negotiating nodes.In this paper, we present the Guarantee and Reservation Protocol (GR Protocol), which can be used by dynamic LD algorithms employing batch task assignments. Principally, the GR Protocol is used for obtaining the mutual agreement between a task sender and a receiver on the appropriate task batch size.The key in this protocol is the virtual workload model which ensures coordinated task dispersements within the distributed systems, while at the same time avoids processor thrashing [4].That is, to avoid situations when a receiver node is flooded with excessive incoming tasks from multiple sender nodes and. thus becomes overloaded.

show abstract

An adaptive algorithm for resolving processor thrashing in load distribution

Cited by 12 publications

References 4 publications

A Negotiation Protocol for Dynamic Load Distribution Using Batch Task Assignments

A Negotiation Protocol for Dynamic Load Distribution Using Batch Task Assignments

Dynamic load distribution using anti-tasks and load state vectors

A negotiation protocol for batch task assignments in dynamic load distribution

Contact Info

Product

Resources

About