www'cs'rmit'edu'aueCDS

1
Dynamic Task Assignment Load Index for
Geographically Distributed Web Services
PhD Research Proposal By Dhiah Al-Shammary
Supervised by 1) Prof. Zahir Tari
2)Ass. Prof. Peter Bertok
www.cs.rmit.edu.au/eCDS
2
Presentation Layouts

• Introduction Background
• Significance
• Project Focus
• Methodology
• Conclusions Future Work

www.cs.rmit.edu.au/eCDS
3
Congestion Bottlenecks
1. Introduction Background

• Popular web services often suffer

Figure 1 Flash Crowds
www.cs.rmit.edu.au/eCDS
4

• Suggested Solution

v Mirrored Web Servers
Figure 2 Mirrored Web Servers
Problem (Survive)
(Congestion)
Figure 3 example on-line travel agent service
www.cs.rmit.edu.au/eCDS
5
Figure 4 A Cooperative and Coordinated
Geographically Distributed Web Servers.
www.cs.rmit.edu.au/eCDS
6
2. Significance

• Prevents systems from crashing
• heavy traffic conditions (flash crowds)
• Improves response time
• The proposed system will develop improved
  geographically distributed web server
  technologies
• efficient process of client request

www.cs.rmit.edu.au/eCDS
7
3. Project Focus

• To develop decentralised cooperative CSDNs
• scale deliver services to
  end-users in a timely
• and reliable manner.

• To produce new theoretical models
• analyse the complexity of
  network congestion problems
• deal with requirements
  (heavy-tailed service demands.)

• To produce a dynamic approach for estimating
  bandwidth and
• network latency

www.cs.rmit.edu.au/eCDS
8
4. Methodology

• Activity 1. Develop a test-bed for evaluation
  of load balancing strategies (OMNET)

• Activity 2. Develop analytical models and
  compare with results from test-bed. Results
  obtained from the test-bed will be used to refine
  the theoretical models for the proposed load
  balancing strategies.

• Activity 3. Develop network parameter
  estimation strategies. Parameter estimation
  strategies will be developed to obtain bandwidth
  and network latency estimations in CSDNs.

www.cs.rmit.edu.au/eCDS
9

• Activity 4. Perform parameter tuning with the
  help of the test-bed. Further optimisations will
  be performed by fine-tuning the parameters used
  by the proposed algorithms in the test-bed
  environment.

• Activity 5. Implement and evaluate the proposed
  strategy in live environment. The proposed
  algorithms and CSDN frameworks will be
  implemented in a modular fashion for integration
  with commercially available tools for Web
  management.

• Activity 6. Writing Thesis In this activity,
  the final thesis will be written in different
  progressive stages focusing on our contributions
  that have been done during the work of this
  project and provide a suitable comparison with
  other works that have been done by other
  researchers.

www.cs.rmit.edu.au/eCDS
10
5. Conclusions Future Work
5.1 Conclusions

• Develop theoretical and computational models for
  co-operative resource sharing for heterogeneous
  and geographically distributed web servers.

• Propose a scalable based system to assist the
  creation of CSDNs (Content/Service Delivery
  Networks) that scale and support sharing of
  resources with other CSDNs.

5.2 Future Work
Implement the proposed system in real time system
make a comparison between the results that we
will get from this system and the first one that
we are working on now.
www.cs.rmit.edu.au/eCDS
11
Thank you very much for your attention
www.cs.rmit.edu.au/eCDS