Homework Assignment 2

W.D.M Dhanushka
080083C

1.)
1. Software Failures
The incidents of software failures on the Web suggest that a large number of non-malicious failures occur during routine
maintenance, software upgrades and system integration. It is unclear whether these failures are mainly due to system
complexity, inadequate testing and/or poor understanding of system dependencies since site owners tend to be vague
about the underlying cause of the failure.
Other significant causes of software failure are system overload, resource exhaustion and complex fault-recovery
routines.
Ex-
• Resource exhaustion
Memory leaks
Thread pool exhausted leading to slow or unresponsive server

• Computational/logic errors
Reference non-existent table in database
Pricing errors

• System Overload
System overload can lead to resource contention and unanticipated interactions between other software processes
running on the same machine, e.g., synchronization errors such as race conditions.

2. Human/Operator Errors
We classified operator errors into three categories.

1. Configuration errors
Occur when operators define incorrect settings for system parameters, e.g., specifying an insufficient number
threads to service user requests to the web server.
Ex-
System reset to default configuration by mistake
Accidentally corrupt configuration file
Omit necessary services from startup script
2. Procedural errors
Occur when the operator either omits an action or executes an incorrect action during system maintenance.
Table 2 describes examples of operator errors.
Ex-
Failure to back up database
Restored the wrong backup
Forgot to restart web server
3. Miscellaneous accidents
Ex-
Equipment trauma caused by drop or fall
Accidentally disconnect power supply

3. Hardware and Environmental Failures

Hardware failures can occur for several reasons, e.g., wear and tear of mechanical parts, design flaws, loose wiring.

• Hardware failures
Ex-
Device driver failures
I/O errors, e.g., hard disk failures (see database media failures)
Memory parity errors

• Environmental failures
Ex-
Power outages
Overheating
High humidity
Natural disasters, e.g., hurricanes, floods
4. Security Violations
This report focuses on non-malicious failures. However, we included examples of failures due to security violations for
the sake of completeness. The list below highlights some of the common security violations that occur in websites.
• Password disclosures
• Denial of Service (DOS) attacks
• Worms and viruses
• Browser vulnerabilities
• Authentication failures
• Theft

2.)
The manifestation of the failure defines as the user-visible effects of a failure.
Manifestation failures can be categorized as follows.
1. Partial or entire site unavailable
2. System exceptions and access violations
3. Incorrect results
4. Data loss or corruption
5. Performance slowdowns

3.)
• Operator errors
• Software failures

4.)
1. University of Michigan’s ICPSR web server failure in July 2002 because web logs grew too large due to
a significant increase in traffic:
According to validation technique it ensures that the product actually meets the user's needs. So in first place
if they check the system for large set of users, they could understand that their system requires to be extended for large
amount of capacity.

2. Failed updates to the basecamphq.com database in February 2005:

This failure occurs due to logical failure in handling of a statement. So if they performed a validation process
for their system they may found application bugs like,
Flawed query may lead to incorrect computations
Inadequate table space allocation for the user/operation

3. Zopewiki.org memory leak in July 2004:

This kind of error occurs because of web server & Application server failures. So if they performed a
validation process for their system performance they may found errors like memory leakage.

4. Bug in the ad server at Double-Click which caused an 80% drop in ad delivery in August 2000:
This is a kind of runtime error. It cannot be detected in validation process.

5. Pricing error at Amazon's UK site which listed a pocket PC at under $12:

This kind of errors occur due to computational & logical errors. If they validate system for high sales volumes
this failure might have noticed.