Talk:Fault-tolerant system

Pair and a spare definition

My understanding of 'pare and a spare' is not what is given in this article. I was taught that 'pair and a spare' was two components operating in parallel for redundancy, with a third unit (the 'spare') operating off-line in case one of the two primaries breaks down. This definition needs to be verified. 147.240.236.9 20:34, 26 June 2006 (UTC)

Damage tolerant design is different from fault-tolerant systems, perhaps being a sub-category. It is specific to structural/mechanical analysis involving imperfections in materials.

Graceful degradation

I think graceful degradation is not quite the same as fault tolerance. While the latter is focused on the system and the availability of the service, the former is focused on the user and the quality of the service he or she gets from the system [1].

The focus shift leads to a shift in the requirement analysis: a fault-tolerant system might take longer to respond or ask for a page reload. When my SMTP server reschedule delivery of a message, it's fault tolerance. When it leaves a message in my inbox saying that my e-mail still hasn't been delivered after four hours, it has nothing to do with keeping the system operating. This program had extra code written in it to ensure some level of satisfaction to the user in the event it cannot fulfill their request.

Where the two concepts meet is exactly when graceful degradation of part of a system makes if fault-tolerant thus enabling users to complete their tasks. [2]

Still I think the distinction is important and graceful degradation should not redirect here. TiagoSilveira (talk) 13:47, 14 February 2009 (UTC)

I agree. StevenBlack (talk) 13:03, 20 November 2009 (UTC)

Fault Tolerance in Software Systems by Design Diversity

• http://sunnyday.mit.edu/papers/nver-tse.pdf —Preceding unsigned comment added by 82.83.74.165 (talk) 22:55, 24 January 2008 (UTC)

Antialiasing

I do not understand the example of graceful degradation. In fact I do not even know where the "top" images end and the "bottom" images start, or what the arrows mean. I think the figure is a good try but needs a better explanation. 129.27.152.126 (talk) 09:42, 27 May 2009 (UTC)

I concur. There are several weaknesses with the example:

1. The example needs at least some knowledge about computer images to even begin to understand what is shown. This knowledge is probably not that common.

2. The use of another, unrelated term from the field of computer imagery ("Antialiasing") is misleading.

3. The explanation of the picture is a mess: What is the center image of a 2x2 grid? What is the center top? Center bottom? Which image is added to which other image? What is the relevance of the colors? What is the relevance of the arrows?

4. If I understand the picture correctly (I am still not sure about this, after reading the explanation three times!), the images in the (vertical) center are added to the images with the arrows and the results are shown in the outmost images. If this is the case, then the flow of reading the picture from inside to the outside is very unusual to English speaking readers. It should be something from top to bottom or left to right. Or at the very least alway in the same direction, not bottom to top first and then top to bottom last.

5. Is there some invisible information in the all black image? (Image on the right, third subimage from the top)

I am sorry I cannot correct the image myself. Even after trying really hard to understand the picture I am still not 100% sure what the point of the picture is. I actually do know how transparency works and I am not THAT stupid and I still do not get it. An illustrating example must not be this inaccessible. If this is to be the example illustration for the article it needs some serious reworking. I would even say a completely different illustration would probably be better (because of my first point) but do not have a better idea off the top of my head. — Preceding unsigned comment added by 134.93.136.193 (talk) 11:41, 22 January 2013 (UTC)

Merge proposal

I propose to merge Fault-tolerant system and Fault-tolerant design. I propose the name for the resulting article be Fault tolerance (currently a redirect to Fault-tolerant design). Neither article is particularly long. It doesn't seem like it would be a problem to cover systems and the design of those systems in the same article. I think this would benefit readers. -—Kvng 15:09, 4 February 2013 (UTC)

I would have to concur. I personally attempted to search for Fault tolerance and expected to receive a page that covered what is separated between Fault-tolerant system and Fault-tolerant design. This merge would have saved me some time and convenience. Khifler (talk) 22:19, 9 May 2013 (UTC)

 Done ~KvnG 03:39, 22 November 2013 (UTC)