What does "Robust" mean?

re: What does "Robust" mean?

Mike Dimmick — Fri, 28 Jan 2005 22:10:00 GMT

"So the robustness criteria for Exchange (or any other email system) includes the users data, while the robustness criteria for NTFS doesn't."

At least, not at the moment. I recall that Longhorn (at least, as announced at the 2003 PDC) will feature Transactional NTFS. Is this still slated for Longhorn, and will it protect simple file writes automatically? Obviously atomic updates to multiple files will need additional coding, but what can we expect, if anything, for existing applications?

re: What does "Robust" mean?

Larry Osterman — Fri, 28 Jan 2005 22:17:00 GMT

Mike, that may be true, I don't know. I do know that if it's enabled, it won't be enabled by default - the performance cost of transacting application data writes is prohibitive (network file copies would be twice as slow, for example).

re: What does "Robust" mean?

kip — Sat, 29 Jan 2005 18:55:00 GMT

This is just the best blog. I'm always happy when there's new Larry Osterman to read.

Here's my meager attempt at a "robust" definition:

A program is robust if, by design, it protects critical data and critical access, provides fail-safe operation, and allows for system recovery when the unexpected does occur.

Feel free to pick it apart. I'd be honored.

re: What does "Robust" mean?

Jim Lyon — Sun, 30 Jan 2005 06:31:00 GMT

How about this: A program (or machine) is robust if it does not exhibit failure modes that are costly.

For the ATM, screwing up the customer's account is high cost; failure to do anything isn't. For the telephone network, failure to connect is high cost.

Also a quibble: In the early 80's I was in the electronic security industry. At that time, it was common for ATM's to fail to account for 1 to 2% of the cash they dispensed. Even then, it was cheaper for banks to eat this 1-2% than to pay human tellers. However, every bank understood that it was very expensive to charge a customer's account without dispensing the money -- reconciliation typically cost tens of dollars, and frequently resulted in the loss of a customer, whose acquisition cost was often about $50.

Larry on robustness

Ben Reichelt's Blog — Sun, 30 Jan 2005 11:19:00 GMT

re: What does "Robust" mean?

Rob — Mon, 31 Jan 2005 10:07:00 GMT

This all sounds right and standard for a sensible definition of robustness.

A quibble of my own: where on Earth do you come from that thinks this is a grammatically meaningful sentence:

<blockquote>The team distributed mugs with "INDUSTRIAL STRENGTH" on them (to indicate that NT, unlike previous MS operating systems) needed to be robust enough to work in mission critical environments.</blockquote>

Some mad parenthesis there :)

re: What does "Robust" mean?

Petr Kadlec — Mon, 31 Jan 2005 12:51:00 GMT

Interesting article. My understanding of "Robust" has always been substantially different than your definition. It seems to me that a program that does not cause problems "when operating in its expected use scenarios" is merely _correct_. I mean, if a program fails in its expected use scenario, it cannot be considered as a valid solution at all.

To me, robustness is a virtue proved by "failing gracefully" (not killing anyone around etc.) even when the conditions have not been expected. The response to the problems that were unexpected is the key definition to me. (Well, obviously, you cannot create a system that would be able to handle anything, so the conditions have to be "expected" at least in some generic way, but nonetheless I feel they shouldn't be named "expected scenarios".)

re: What does "Robust" mean?

G. Man — Mon, 31 Jan 2005 14:56:00 GMT

I think the real problem with your definition is that you seem to use "unexpected behavior" and "failure" interchangeably... I think they are 2 different things.

re: What does "Robust" mean?

Larry Osterman — Mon, 31 Jan 2005 16:43:00 GMT

Rob: That's because I don't have an editor to catch my stupid mistakes.

Petr: That's why I have the second sentence (about recovering without lost data). You're right that it's "merely _correct_", and a correct program _should_ be robust.

As I said in the beginning, everyone likes to use the word "Robust" (or "Reliable") but the basic problem here is that the word is meaningless. Is a phone network more robust (or reliable) than an ATM because it's always available?

Robustness is clearly a desirable feature in a system, but how do you know if your system _is_ robust?

re: What does "Robust" mean?

Frank Hamersley — Sun, 06 Feb 2005 13:38:00 GMT

Your definition is most illuminating given your (and your employers) position within the IT industry.

The distinction of "operating in its expected use scenarios" is far too weak IMO and experiance oft reinforces this belief.

Truly robust programmes manage unexpected use cases as well as the (apparently) impossible ones.

Had this ethic permeated the industry some time ago I suspect the times would not be quite as exciting or depressing depending (respectively ) whether you are observing or afflicted with the manic exploit riddled jungle we find ourselves in.

Personally I don't think too many "real" programmers are willingly complict in this but the expediencies driven by Marketers to ship now, fix later and the ultra-rationalist vesting mentat make it difficult for ppl to deliver on such principles.

Frank.

Words I dislike - Policy

Larry Osterman's WebLog — Wed, 27 Jul 2005 18:40:02 GMT

A while back, I wrote about how I disliked the word "Robustness" because it's meaning was so vague. ...

Larry Osterman s WebLog What does Robust mean | Paid Surveys

Larry Osterman s WebLog What does Robust mean | Paid Surveys — Sat, 30 May 2009 13:15:54 GMT

PingBack from http://paidsurveyshub.info/story.php?title=larry-osterman-s-weblog-what-does-robust-mean

Larry Osterman s WebLog What does Robust mean | work from home

Larry Osterman s WebLog What does Robust mean | work from home — Tue, 16 Jun 2009 16:17:18 GMT

PingBack from http://workfromhomecareer.info/story.php?id=22107

Larry Osterman s WebLog What does Robust mean | debt solutions

Larry Osterman s WebLog What does Robust mean | debt solutions — Fri, 19 Jun 2009 21:22:01 GMT

PingBack from http://debtsolutionsnow.info/story.php?id=929