Workload Management and Scheduling

Hasso Plattner

doi:10.1007/978-3-642-36524-9_22

A Course in In-Memory Data Management pp 149-152 | Cite as

Workload Management and Scheduling

Authors
Authors and affiliations

Hasso PlattnerEmail author

Chapter

First Online: 23 April 2013

1.1k Downloads

Abstract

One of the most important factors that determine the usability for an application is response time. Psychological studies show that the acceptable maximum application response time for a human is about three seconds. After three seconds, the user might loose concentration and does something else. Once the application has finished its processing and is ready for the next human input, the user has to focus on the application again. These context switches are extremely expensive as they constantly interrupt the user and lead to being unproductive.

This is a preview of subscription content, to check access.

Self Test Questions

1.
Resource Conflicts
Which three hardware resources are usually taken into account by the scheduler in a distributed in-memory database setup?
1. (a)
  
  CPU processing power, main memory, network bandwidth
2. (b)
  
  Main memory, disk, tape drive
3. (c)
  
  CPU processing power, graphics card, monitor
4. (d)
  
  Network bandwidth, power supply unit, main memory.
2.
Workload Management Scheduling Strategy
Why does a complex workload scheduling strategy might have disadvantages in comparison to a simple resource allocation based on heuristics or a uniform distribution, e.g. Round Robin?
1. (a)
  
  The execution of a scheduling strategy itself consumes more resources than a simplistic scheduling approach. A strategy is usually optimized for a certain workload—if this workload changes abruptly, the scheduling strategy might perform worse than a uniform distribution
2. (b)
  
  Heuristics are always better than complex scheduling strategies
3. (c)
  
  A scheduling strategy is based on general workloads and thus might not reach the best performance for specific workloads compared to heuristics or a uniform distribution, while its application is cheap
4. (d)
  
  Round-Robin is usually the best scheduling strategy.
3.
Analytical Queries in Workload Management
Analytical queries typically are \(\ldots \)
1. (a)
  
  long running with soft time constraints
2. (b)
  
  short running with soft time constraints
3. (c)
  
  short running with strict time constraints
4. (d)
  
  long running with strict time constraints.
4.
Query Response Times
Query response times \(\ldots \)
1. (a)
  
  can be increased so the user can do as many tasks as possible in parallel because context switches are cheap
2. (b)
  
  have to be as short as possible, so the user stays focused at the task at hand
3. (c)
  
  should never be decreased as users are unfamiliar with such system behavior and can become frustrated
4. (d)
  
  have no impact on a users work behavior.

Reference

[Pla11]
H. Plattner, SanssouciDB: an in-memory database for mixed-workload processing, in BTW (Springer, Berlin, 2011), pp. 2–21 Google Scholar

Copyright information

Authors and Affiliations

Hasso Plattner
- 1
Email author

1.Hasso Plattner InstitutePotsdamGermany

Workload Management and Scheduling

Abstract

Reference

Copyright information

Authors and Affiliations

Personalised recommendations

Cite chapter