# Task Assignment in a Server Farm with Switching Delays and General Energy-Aware Cost Structure

@inproceedings{Hyyti2013TaskAI, title={Task Assignment in a Server Farm with Switching Delays and General Energy-Aware Cost Structure}, author={Esa Hyyti{\"a} and Rhonda Righter and Samuli Aalto}, year={2013} }

We consider the task assignment problem to parallel servers with switching delay, where servers can be switched off to save energy. However, switching a server back on involves a constant server-specific setup delay. We will use one step of policy iteration from a starting policy such as Bernoulli splitting, in order to derive efficient task assignment (dispatching) policies that minimize the long-run average cost. To evaluate our starting policy, we first analyze a single work-conserving M/G/1… Expand

#### Supplemental Presentations

#### Figures, Tables, and Topics from this paper

#### 2 Citations

State of the Art and Research Challenges in the Area of Autonomous Control for a Reliable Internet of Services

- Computer Science
- Autonomous Control for a Reliable Internet of Services
- 2018

The goal of this book chapter is to first analyze the state-of-the-art in the area of autonomous control for a reliable IoS and then to identify the main research challenges within it. Expand

Autonomous Control for a Reliable Internet of Services

- Computer Science
- Lecture Notes in Computer Science
- 2018

Traditional networks are transformed to enable full integrationof heterogeneous hardware and software functions, that are configuredat runtime, with minimal time to market, and are provided to thei… Expand

#### References

SHOWING 1-10 OF 57 REFERENCES

Energy-aware dispatching in parallel queues with on-off energy consumption

- Computer Science
- 30th IEEE International Performance Computing and Communications Conference
- 2011

This work devise efficient dispatching heuristics based on the first policy iteration procedure of Markov Decision Processes to minimize a weighted sum of delay and energy consumption in a dynamic dispatching problem where jobs are assigned upon arrival into parallel queues. Expand

Dispatching problem with fixed size jobs and processor sharing discipline

- Computer Science
- 2011 23rd International Teletraffic Congress (ITC)
- 2011

A closed form expression is derived for the so-called size-aware relative value of state, which sums up the deviation from the average rate at which sojourn times are accumulated in the infinite time horizon, which can be applied in numerous situations. Expand

Optimal state-free, size-aware dispatching for heterogeneous M/G/-type systems

- Computer Science
- Perform. Evaluation
- 2005

It is proved that the optimal strategy for systems with identical servers assigns a non-overlapping interval range of job sizes to each server, and that when server processing speeds differ, it is necessary to assign each server a distinct set of intervals ofJob sizes in order to minimize expected waiting or response times. Expand

Minimizing slowdown in heterogeneous size-aware dispatching systems

- Computer Science
- SIGMETRICS '12
- 2012

The main contribution of this paper is to show the optimality of SPTP with respect to slowdown in a single server queue under Poisson arrivals, and to derive the so-called size-aware value functions for M/G/1-FIFO/LifO/SPTP with general holding costs to derive efficient dispatching policies so as to minimize the mean slowdown in an heterogeneous server system. Expand

M/M/1-PS queue and size-aware task assignment

- Computer Science
- Perform. Evaluation
- 2011

Numerically, it is shown that for the exponentially distributed job sizes the myopic approach, ignoring the future arrivals, yields an efficient and robust policy when compared to other heuristics, however, in the case of highly asymmetric service rates, an FPI based policy outperforms it. Expand

Optimality of D-Policies for an M/G/1 Queue with a Removable Server

- Computer Science
- Queueing Syst. Theory Appl.
- 2002

It is proved that D-policies are optimal for the average cost per unit time criterion, which means that for this criterion there is an optimal policy that either runs the server all the time or switches the server off when the system becomes empty and switches it on when the workload reaches or exceeds some threshold D. Expand

Surprising results on task assignment in server farms with high-variability workloads

- Computer Science
- SIGMETRICS '09
- 2009

This paper investigates the performance of task assignment policies for server farms, as the variability of job sizes (service demands) approaches infinity, and shows SITA to be inferior to the much simpler greedy policy, Least-Work-Left (LWL), for certain common job-size distributions. Expand

On Choosing a Task Assignment Policy for a Distributed Server System

- Computer Science
- J. Parallel Distributed Comput.
- 1999

This work explores the influence of task size variability on which task assignment policy is best and uses the resulting observations to argue in favor of a specific size-based policy, SITA-E, that shows very good performance for realistic task size distributions. Expand

Server farms with setup costs

- Computer Science
- Perform. Evaluation
- 2010

This paper derives the first closed-form solutions and approximations for the mean response time and mean power consumption in server farms with setup costs, and applies this analysis to data centers, where both responseTime and power consumption are key metrics. Expand

A heuristic rule for routing customers to parallel servers

- Computer Science
- 1997

A heuristic solution method yielding a good suboptimal rule will be given for the case of server groups with different and generally distributed service times based on a decomposition approach and first principles from Markov decision theory. Expand