1

Robustness and Reliability

Kupat Tahu Presentation

CHAPTER-3
ROBUSTNESS AND RELIABILITY
Objectives:

• To discuss requirements of robustness and reliability

• To discuss categories of Applications
• To understand failure scenarios and reliability challenges

Learning Outcomes:
At the end of this lesson the students will be able to understand:

• Robustness and reliability and their requirements

• Different categories of applications
• Reliability challenges
PREVIOUS LECTURE CONTENTS
• SOA-BASED ARCHITECTURE
• API-ORIENTED ARCHITECTURE
• Resource Management
• IoT data management and analytics
LECTURE CONTENTS
• Reliability requirement
• Categories of applications
• Failure scenarios
• Reliability challenges
 5
INTERNET OF THINGS—ROBUSTNESS AND RELIABILITY

• Building a reliable computing system has always been an important

requirement for the business and
the scientific community. By the term reliability, we mean how long a
system can operate without any failure.
• Another closely related quality attribute, called availability
• Availability has been one of the most important quality attributes to
measure the extent of uninterrupted service that a distributed and
more recently a cloud-based system provides.
 6
Requirement

Availability: percentage of time that a system is operational to the

user
Reliability: how long a system can operate without any failure
MTTF: Mean Time To Failure
MTTR: Mean Time To Repair
• Recovery oriented computing(ROC) MTTR as small as
possible

Availability A can be computed as:

MTTF is the long-run average time to failure where MTTR denotes the average
time the system takes to be operational again after a failure.
Serviceability: ease with which the deployed system can be
repaired and brought back to operation .
 Zero repair time
 7

MTTR

MTTR often includes the time to do the following:

1. Notify the relevant repair workers.

2. Diagnose the problem.
3. Repair the problem.
4. In the case of code, fix the bugs.
5. In the case of code test to make sure repairs haven’t affected other parts of the program.
6. In the case of physical equipment, allow the equipment/devices to cool down.
7. In the case of physical equipment, reassemble the device and make it ready for use again.
8. Finally, set up and test the device.
 8
IoT CHARACTERISTICS AND RELIABILITY ISSUES

• With the advancement in infrastructure and wireless communication, proliferation of

new communication aware devices of various form factors, and with the introduction of
cloud computing paradigm, Internet of Things-based applications are emerging.
• For instance,
• Smart, network accessible cameras can be placed at strategic locations in a cluster of
buildings or on streets,
• Smart meters may be installed in a power-grid,
• Tiny embedded devices can be used for health monitoring,
• Vehicles in a city can be equipped with GPS-based sensors, and
• Static wireless sensors can be embedded in modern appliances like a television or a
refrigerator.
These network-enabled devices can run distributed processes, which in turn can
coordinate, exchange data, and take critical decisions in real time.
IoT ARCHITECTURE IN BRIEF 9

Though there is no consensus of what the standard architecture of an IoT-based system should be, we
find a general adoption of the following IoT architecture reference model.
1. Device layer:
2. Communication
layer:
3. Application layer:
 10
Categories of Applications

1. Zero tolerance:
• When an IoT system is placed in a mission critical scenario, network aware devices monitor the
health of the patient, or a network-enabled pacemaker device that interacts with a larger health-
care platform
• System components do not tolerate any failure during its mission time.
• MTTF (which impacts reliability) for these components should be strictly greater than the
mission time.
• MTTR for such a system should be close to zero during the mission time.
 11
Categories of Applications

2. Restartable:
• IoT system can tolerate a faulty component or even the entire system (though undesirable) to restart without
any catastrophic impact.
• For instance, an IoT system for urban transport, having components embedded in vehicles, can afford to
restart, if the embedded component fails.
• Safety Hazards, Economic Losses, Service Disruptions, Legal and Regulatory Consequences
• Here, more than MTTF being high, the goal is to make MTTR as small as possible.
 12
Categories of Applications

3. Error Tolerant:
• The nature of the application is such that a part of the system can
tolerate the erroneous input for some time, within the user-defined
safety limit before getting it fixed.
• For instance, an unmanned surveillance system providing real-time
routine information of an agricultural land, can afford to send
poor/incorrect data before it is rectified
 13
FAILURE SCENARIOS

IoT-based system becomes unreliable or unavailable when the system either

fails to respond to a request or provides an unexpected, incorrect service.

A service failure happens when faults are not handled properly

Faults are categorized as:

Development fault
Infrastructure fault
Interaction faults
 14

• Development fault
• Induced by incorrect implementation of the software

• Infrastructure Fault
• Physical deterioration
• External environment
• Processors with limited functionality
• Limited compute due to battery

Reliability and availability of the system will depend on the extent to which these
devices can withstand these unexpected scenarios
 15

• Interaction Fault
• Entire network or the communication components can fail
• Impedance mismatch due to the heterogeneity of the devices in
databases, impedance mismatch often arises when trying to map data
between different types of databases, such as relational databases and
object-oriented databases..
• Unexpected workload coming from various IoT components
 16
RELIABILITY CHALLENGES
• Making service available to user
• delivering the service to the user, not just surviving through a failure.
• “user perceived availability”

• Serviceability of IoT System

• System is operational for a longer duration without any repair
• Traditional maintenance may not be feasible for an IoT system

• Device Level Reliability

• For sensor devices, the environmental condition can result in a bit error.
• Devices deployed in medical monitoring and diagnostic systems need to
be safety critical.
• Devices used in fire safety scenarios need to be zero tolerant
 17

• Reliability at Network Level

• Network must be able to configure itself as environmental conditions or
components in the network itself change so that the information can always be
transmitted from one application to the other reliably
• Signals from a sensor device interfere with another and make the links
unreliable.
• Power-saving as UDP and as reliable as TCP.

• Privacy and reliability

• Identity management, and proving identity on-demand has been
considered an important mechanism to ensure the authenticity of the
communicating parties
• IoT-based vehicle management system
 18

• Interoperability of devices
• Heterogeneous devices cannot exchange information due to the lack of
standardization.

• Reliability issues due to energy constraint

• System performance can be impacted due to the limited power supply.
• Infrastructure ensures both reliability and low energy consumption
References
• Internet of Things: Principles and Paradigms, Rajkumar Buyya,Amir Vahid Dastjerdi, ISBN: 978-
0-12-805395-9

19