Dependable Computing

1
Dependable Computing

• Kshitish Chitnis Developing Effective Cluster
  Connectivity
• Mukesh Thakur Modeling Cost Based Availability
  and Reliability
• Lixin Shen Reliability Modeling using
  Distributed Computing Architectures.
• Hertong Song Reliability Modeling using
  Distributed Architectures Infrastructure
  Development.

2
Project Title

• Developing Effective Cluster Connectivity

Kshitish Chitnis
3
Problem Statement

• Devise an effective cluster computing solution
  using the USB technology protocol support.
• We plan to create a device driver that helps
  manage a Linux Cluster maintaining
  high-availability and reliability in the cluster.

4
Overview

• High-Availability and Reliability issues in
  cluster computing.
• The device driver will be developed to support
  cluster connectivity. Various configurations will
  be experimented with, to form different redundant
  network topologies, e.g. simulated token ring,
  hyper-cube, star, tree.
• An optimum model will be decided upon.

5
Benefits

• Our expected result will be for better
  understanding improving system reliability via an
  alternative redundant communication path and an
  enhanced clustering tool for HA.

Back
6
Project Title

• Modeling Cost Based Availability and Reliability

Mukesh Thakur
7
Problem Statement

• Complex systems with dependencies cannot be
  easily characterized neither by fault trees nor
  by RBD solutions. Hence we intend to use Markov
  models to perform the dependency analysis.
• Design tools that will help us pin point problem
  areas on modeled systems and perform a cost based
  analysis to obtain a reliability estimation of
  the system.

8
Overview

• Being able to perform statistical analysis of
  modeled systems can help us design more fault
  tolerant and reliable systems.
• Cost can be a prohibiting factor and hence being
  able to model based on cost considerations can be
  a handy tool for the system designer.

9
Benefits

• A tool that will enable system designers and
  architects to estimate the target system
  reliability and perform cost based calculation
  while implementing an optimal system.

Back
10
Project Title

• Dependability and Reliability Modeling using
  Distributed Computing Architectures.

Lixin Shen
11
Problem Statement

• Distributed computing architectures can be
  difficult to model due to the complexities
  involved and the number of dependencies.
• There can be various problem domains under
  consideration and we would want to investigate
  appropriate reliability modeling techniques for
  our problem domains.

12
Overview

• There are various reliability modeling tools that
  available, we plan to study them in detail and
  understand the strengths and limitations of the
  existing tools.
• This will enable us to design a better tool that
  can include features from the existing tools so
  as to avoid reinventing the wheel and thereby
  also create a tool that is a significant
  enhancement over the existing prototypes.

13
Benefits

• A tool that will enable system designers and
  architects to estimate the target system
  reliability and evaluate potential points of
  failure in the system typically targeting
  distributed computing architectures.

Back
14
Project Title

• Reliability Modeling using Distributed
  Architectures Infrastructure Development.

Hertong Song
15
Problem Statement
16
Overview
17
Benefits
Back