Resolving QoS Policy Configuration Challenges for PublishSubscribe Middleware Platforms

1
Resolving QoS Policy Configuration Challenges for
Publish/Subscribe Middleware Platforms

• AFRL JBI PI Meeting

2
Outline

• Motivating Example
• QoS Policy Configuration Challenges
• Evaluating Common Solutions
• DQML Overview

3
Motivating Example Unmanned Aerial Vehicle (UAV)

• UAV Desirable Data Flows
• Storage and dissemination of data for late
  arriving subscribers (e.g., other aircraft and
  satellites coming into range)

• Reliable communication of critical data (e.g.,
  sensor data to satellite to ground station)

• Prioritization of data delivery for
  mission-critical or high value data (e.g.,
  time-sensitive targets)

• Ordered and/or grouped data dissemination to
  ensure ordering and appropriate level of
  granularity

4
QoS Policy Configuration Challenges (1 of 3)

• QoS Policy Compatibility
• QoS policies for the communicating entities must
  be compatible between whats requested and offered

• For example, if subscriber requests reliable data
  transfer the publisher can not offer best effort
  data transfer

5
QoS Policy Configuration Challenges (2 of 3)

• QoS Policy Consistency
• QoS policies for any one entity must be
  consistent with each other

• For example, deadline period must be gt minimum
  separation for time based filter

6
QoS Policy Configuration Challenges (3 of 3)

• Accurate QoS Policy Configuration Transformation
• Propagating QoS policy configuration through
  development stages

• For example, transforming design of QoS
  configuration into application code

7
Evaluating Common Solutions

• Three Common Solutions to Challenges
• Point-based solutions

• Patterns-based solutions

• Domain specific modeling language (DSML) based
  solutions

• Categorized by
• Robust vs. non-robust solution documentation
• Robust vs. non-robust solution implementation

8
Point-based Solutions

• Focused on particular solution/application
• QoS policy configuration is developed
• Configuration is designed

Code
Design

• Configuration is implemented

• Application is compiled run

• Feedback is gathered evaluated

• Process is repeated as necessary

Test
Evaluate

• Cons
• Non-robust solution documentation
• Non-robust solution implementation

• Pros
• Low overhead

9
Patterns-based Solutions
Continuous Data Pattern

• Enables codification of configuration expertise

• Documented use of QoS policies for
• Dataflow management
• Dataflow prioritization
• Dataflow shaping

• Cons
• No help with solution implementation (i.e.,
  non-robust)

• Pros
• Reuse of expert configuration design knowledge
  (i.e., robust design)

Application B
Application C
Application A
10
DSML-based Solutions
Metamodel

• Uses domain specific terminology and constructs

Application model

• Metamodel created in metamodeling tool

• Cons
• Learning curve/training overhead

• Pros
• Robust solution documentation (via metamodel
  constraints)
• Robust solution implementation (via interpreters)

QoS Config
QoS Config
Application C
Application B
Application A
11
DDS QoS Modeling Language (DQML 1 of 2)

• Models relevant DDS entities

• Models DDS QoS polices

• Models relationships between entities and QoS
  policies

12
DDS QoS Modeling Language (DQML 2 of 2)

• Supports QoS compatibility and consistency
  constraint checking

• Generates implementation artifacts (currently for
  DDS Benchmarking Environment (DBE))

DBE Interpreter
DBE