Dependability

1
Dependability

• How much can you trust your program?
• How important is it that you trust your program?
• Note Usefulness and trustworthiness are not the
  same thing.

2
Dependability

• Availability
• How likely that system will be up and running?

• Reliability
• How likely that system behaves correctly?

• Safety
• How likely is system to cause damage?

• Security
• How likely can system defend against accidental
  or deliberate intrusion?

3
Cost/dependability curve
C
o
s
t
Dependability
L
o
w
M
e
d
i
u
m
H
i
g
h
V
e
r
y
U
l
t
r
a
-
h
i
g
h
h
i
g
h
4
Dependability vs. Performance

• High levels of dependability often achieved at
  expense of performance
• Justification
• Undependable critical systems wont be used
• Cost of system failure very high
• Data integrity/reliability very important
• Dependability hard to retrofit
• Possible to compensate for performance deficit

5
Critical Systems

• Systems where failures can result in physical
  damage, threats to life, significant economic
  damage
• Safety-critical systems
• Mission-critical systems
• Business-critical systems
• Software only one link in the chain
• hardware
• human operator

6
Availability and Reliability

• Both can be quantified as probabilities
• Reliability assumes availability, but might have
  different requirements, owing to
• reboot time
• perceptions

7
Availability and Reliability

• ReliabilityThe probability of failure-free
  operation over a specified time in a given
  environment for a specific purpose
• AvailabilityThe probability that a system, at a
  point in time, will be operational and able to
  deliver the requested services

8
Reliability and context

• Perception of reliability influenced by
• patterns of use
• human perceptions
• situations where failures occur

9
Reliability Terminology

• System failure
• when system doesnt deliver user-expected service
• System error
• when system doesnt behave to spec
• System fault
• incorrect system state
• Human error or mistake
• source of error is human behavior

10
Improving Reliability

• Fault Avoidance
• Conservative programming idioms
• Fault Detection and Removal
• Rigorous testing
• Fault Tolerance
• Designing system to recover gracefully from
  runtime faults

11
Number of Bugs ! Unreliability

• Removing 60 of produce defects improved
  reliability only 3 Mills
• Its how often the bugs are tickled that matters
• and unfortunately, this could relate to user
  behavior

12
Safety and Security

• Harder to quantify than reliability or
  availability
• Process strategies
• avoidance
• detection and removal
• limitation of influence