Consistency In Continuous Distributed Interactive Applications

1
Consistency In Continuous Distributed
Interactive Applications
Dr. ZHOU Suiping School of Computer
Engineering Nanyang Technological University
2
Outline
3
Why (do we address the consistency issue)?

• Distributed Interactive Applications (DIAs) have
  been
• widely used and studied in recently years.
• DIS, Multi-user games, cooperative editing
  systems,etc

• Two defining properties of DIAs

• Various inconsistency may occur which may
  greatly
• degrade the performance of the application.

4
How (do we address the consistency issue)?

• Classification of applications

We focus on continuous applications.

• A consistency model

Provide a framework for analysis and maintenance

• Consistency maintenance

5
A Consistency Model
Simulated world
site A
site B
Physical world
Transmission delay
Action
Action
Time-stamped message
Transmission delay
Various kinds of inconsistency
6
A Consistency Model
Note

• Inconsistency is ubiquitous
• Different phenomena

7
A consistency Model
8
A Consistency Model
Definition 1
Suppose E is a moving entity in an application,
and is its simulated position at site
at wall-clock time t, where n is the number
of sites in the application. If
, the application is called absolutely
consistent.
9
A Consistency Model
How big an inconsistency is considered
as an inconsistency ?
10
A Consistency Model
In order to evaluate the time-space inconsistency
property of an application, we define
Relative time-space inconsistency
as
Not discernible
Not Obvious
Obvious
11
A Consistency Model
12
The Computation of
It is important to estimate the relative
time-space inconsistency!
For applications have quasi-periodical behaviors,
it is possible to estimate
13
The Computation of
14
Consistency Maintenance
Consistency
Responsiveness
Global issue - Relationship among different
local views
Local issue -
15
Consistency Maintenance
16
Consistency Maintenance
The short period after an action is taken on an
entity.
Critical Situation
Why critical?
17
Future Work

• A set of algorithms for consistency maintenance
• based on critical situation slow-down.

• Further relax our consistency model for specific
  
• application, such as turn-based applications.