When Algorithms Go Wrong

An example of when an arithmetic operation led to unexpected behavior in computer usage and design was with the European Space Agency’s Ariane-5 launcher. The Ariane-5 launcher’s flight lasted only forty seconds because of an overflow error. According to Le Lann (2002), the reason the Ariane-5 launcher exploded was “during a conversion from a 64-bit floating point number F to a 16-bit signed integer number”. A 16-bit signed integer can only hold integers ranging from -32,768 to 32,767. When the conversion from the 64-bit floating number resulted in a larger integer than the 16-bit signed integer can hold, it created an overflow error.

Unfortunately, the error was not fixed in time because the Ariane –5 launcher had disintegrated shortly after takeoff. Lake (2010) states that this error occurred because a “software handler that could have dealt with the problem had been disabled”. This overflow error could have been avoided by enabling a software handler to work alongside the existing software. Another possible solution would be to not covert the 64-bit number to a 16-bit number in the first place. This solution could have easily stopped the overflow and prevented the mishap. Also, if compatible hardware was used it would have made it so that the 64-bit floating point number would not have had to be converted. Lastly, the test that occurred in which the overflow happened were needed for the previous design, the Ariane-4, but were not needed for the Ariane-5 (Le Lann, 2002). If the features that were not needed for the new design were removed, the error would not have occurred.

References

Le Lann, G. (1997, January). An Analysis of the Ariane 5 Flight 501 Failure - A System Engineering Perspective. Research Gate. Retrieved November 18, 2021, from https://www.researchgate.net/publication/220882405_An_analysis_of_the_Ariane_5_flight_501_failure-a_system_engineering_perspective

Lake, M. (2010, September 09). Epic failures: 11 infamous software bugs. Retrieved November 18, 2021, from https://www.computerworld.com/article/2515483/epic-failures-11-infamoussoftware-bugs.html?page=3

Comments

Peering Points and the Network Application Interface

According to Gibb (2019), “Peering is a method that allows two networks to connect and exchange traffic directly without having to pay a third party to carry traffic across the Internet.” Utilizing a peering point allows users to send a receive data directly to one another without the need to route through other computer networks. Doing so allows for a quicker, more efficient, and safer form of communication. Researching the total number of active Internet Exchange Points (IXP) proved difficult as it was hard to pinpoint an accurate number. However, according to Rosas (2021), “as of January 2021, of the 630 registered IXPs, 229 are in Europe, 126 in North America, 140 in Asia-Pacific, 96 in Latin America and the Caribbean (LAC), and 39 in Africa.” These numbers are constantly changing as new IXPs are added, and some are removed. Finding a definitive number of Internet Service Providers (ISP) globally was also difficult to accomplish. Most sources seem to point to the Nations Encyclop

VLAN Aggregation for Efficient IP Address Allocation

The project I chose to summarize on the Internet Engineering Task Force (IETF) website was RFC: 3069, VLAN Aggregation for Efficient IP Address Allocation. Within this project, the authors point out how inefficiently a Virtual Local Area Network (VLAN) allocates IP addresses along with their proposed solutions. I have also attached a diagram showing how the network would look pertaining to this project. Currently, an IP subnet would be made for each existing customer by understanding how many hosts they currently need and may need in the future. Based on that total number, the IP subnet and gateway address would change according to how many hosts the customer requested. For example, if a customer has indicated that they need ten hosts, and they only use five, the additional five that are not in use cannot be used by another customer. An illustration of this is shown below. The proposed solution to this problem is to allocate IP addresses under the same IP subnet and gateway address uti

Access Lists and Capabilities

To implement an access matrix, we must first understand what it is. An access matrix is a protection model within an operating system consisting of objects and domains. The access matrix determines which processes interact with objects within the domain. Objects within the domain can consist of both hardware and software. The lists below show the advantages and disadvantages of access lists associated with objects, and capabilities with domains. Access lists associated with objects Advantages Corresponds directly to the user’s needs. Easy revocation and review of access. Disadvantages Difficult to determine access rights for a domain. Takes time to search the domain for access rights. Capabilities with domains Advantages Useful for localizing information for a process. Secured against unauthorized access. Disadvantages Inefficient at the revocation of capabilities. Does not correspond directly to the user’s needs. Even though each implementation has its own strengths and w

An Adventure Into Technology

Search This Blog