SQLServerCentral / Article Discussions / Article Discussions by Author / Discuss content posted by Hugo Kornelis / Index types 2 / Latest Posts

RE: Index types 2

L' Eomot Inversé — Mon, 03 Sep 2012 05:07:02 GMT

[quote][b]Alex Fekken (9/3/2012)[/b][hr][quote][b]L' Eomot Inversé (9/2/2012)[/b]Well, the non-clustered index with included columns is an amusing concept, and I'm astounded that so many people (44% now) believe in such a fabulous beast.[/quote]Not only do I believe in them, but I use them a lot as well. :-)[/quote]:blush: :w00t:So do I![quote]If you meant "clustered index with included columns", then I believe in those as well because a clustered index always includes all of its columns. So this should have counted as a correct answer. (The answer wasn't about clustered indexes with an INCLUDE clause in their definition, but clustered indexes with included columns)[/quote]yes, that's what I meant. And to me the question seemed to be about t-sql syntax as well as semantics, so it was indeed about clustered indexes with INCLUDE clause.

RE: Index types 2

Alex Fekken — Mon, 03 Sep 2012 03:51:45 GMT

[quote][b]L' Eomot Inversé (9/2/2012)[/b]Well, the non-clustered index with included columns is an amusing concept, and I'm astounded that so many people (44% now) believe in such a fabulous beast.[/quote]Not only do I believe in them, but I use them a lot as well. :-)If you meant "clustered index with included columns", then I believe in those as well because a clustered index always includes all of its columns. So this should have counted as a correct answer. (The answer wasn't about clustered indexes with an INCLUDE clause in their definition, but clustered indexes with included columns)

RE: Index types 2

L' Eomot Inversé — Sun, 02 Sep 2012 11:31:03 GMT

Very good question.[quote][b]Hugo Kornelis (8/30/2012)[/b][hr]BTW, maybe it's a bit early to say anything about the percentages, but at this point I am mainly surprised at:* 28% wrong answers on the "bitmap on computed" option - I expected this to be lower* 17% wrong answers on the "nonclustered on a view" option - here, I expected a higher failure rate[/quote]Well, the [strike]non-[/strike]clustered index with included columns is an amusing concept, and I'm astounded that so many people (44% now) believe in such a fabulous beast. :-)Maybe people have some vague idea about bitmap indexes because a certain non-MS dbms offers its users this interesting form of mental agony, and that vague idea misleads them into guessing that they exist in MS SQLServer? Anyway, the number getting that one wrong is now down to 22%.I guess these two numbers are a bit closer to your expctations now that they were early on..[quote][b]Hugo Kornelis (8/30/2012)[/b][hr][quote][b]DugyC (8/30/2012)[/b][hr]Had to answer by process of elimination[/quote]Thanks, Dugy!Frankly, I am a bit disappointed that the site moderators included the number of correct answers. I had included a note that there are multiple correct answers without specifying how many, to make it a true test of knowledge - but apprently, including the number of correct answers for questions with more than one answer is now standard practice, applied to all those questions. :([/quote]Well, I looked up the maximum column count for columnstore index anyway, because I knew there was a limit, and wanted to know what the limit was (obviously it was 260 or higher, but I didn't see knowing that as a reason not to go look it up - and I would have looked it up anyway if the count of 4 hadn't been given). I don't think generally that giving a count is a bad thing, although maybe for some questions it may give too much away. Do you really think it does here? (Maybe it saves people some time researching "bitmap" indexes, but if the objective is for people to learn about SQL Server rather than to learn about Mr Ellison's software that is perhaps a useful saving.)edit: eliminate non-

RE: Index types 2

Lokesh Vij — Fri, 31 Aug 2012 00:57:36 GMT

Thanks for the great question Hugo!Kept me researching to figure out the correct answer. In the end, I followed the proecss of eliminitation :-)Very sure that you must have spent lot of time creating such a wonderful question!

RE: Index types 2

mtassin — Thu, 30 Aug 2012 14:01:07 GMT

Glad to be part of the 27% that got it right, but that's only because I had the 4 to go by.I would have probably bypassed the columnstore one otherwise. The rest I was pretty sure on either way, but the columnstore one was more plausible than the bitmap on a computed column one.

RE: Index types 2

Narud — Thu, 30 Aug 2012 12:56:53 GMT

Yes! Proudly I'm part of the 73% that went wrong. :pinch: But I've learned something new, and now I know a little bit more of SQL of what I knew yesterday.Hugo, thanks for this good question.

RE: Index types 2

SQLRNNR — Thu, 30 Aug 2012 10:25:35 GMT

Good question Hugo.

RE: Index types 2

Erav — Thu, 30 Aug 2012 10:17:19 GMT

Great Question..-Erav

RE: Index types 2

EL Jerry — Thu, 30 Aug 2012 09:49:28 GMT

Great question, Hugo, got it wrong because of poor researching, but this is something good to have learned.

RE: Index types 2

sestell1 — Thu, 30 Aug 2012 07:43:51 GMT

Argh, good question Hugo.Sadly, I checked off the ones that weren't valid, and forgot to switch them before hitting submit. :angry:

RE: Index types 2

bitbucket-25253 — Thu, 30 Aug 2012 07:39:33 GMT

I read this [quote]Note that some index types may require additional attributes to be specified; this does not invalidate them for the context of this question. Only index types that include illegal attribute combinations or non existing attributes should be discarded. Also note that this question is not about the exact spelling of the various options[/quote]and immediately realized that it was another question by Hugo Kornelis. So back to BOL, Google and multiple cups of coffee. But once again I learned something ... Mr Kornelis I thank you.

RE: Index types 2

(Bob Brown) — Thu, 30 Aug 2012 07:39:21 GMT

Thanks for the question. Couldn't get it right even with researching it!

RE: Index types 2

paul.knibbs — Thu, 30 Aug 2012 07:34:19 GMT

I'd never heard of a bitmap index, but I knew which four answers were right and so did OK. If it hadn't specified there were four right answers I'd have struggled a bit with the bitmap answer, because it's often quite difficult to find something that says "X is unsupported in SQL Server"! :-P

RE: Index types 2

Dana Medley — Thu, 30 Aug 2012 06:48:20 GMT

[quote][b]Thomas Abraham (8/30/2012)[/b][hr]Thanks for the question Hugo. I should have stretched out before running that hard early in the morning. I think I strained something! :-) I certainly learned something.[/quote]For starters + 1 :-DExcellent question Hugo. Made me think a lot more this morning than I have in a while. I can say without a doubt that I have learned something.

RE: Index types 2

sknox — Thu, 30 Aug 2012 06:14:34 GMT

[quote][b]Hugo Kornelis (8/30/2012)[/b][hr]Thanks, Koen! (EDIT: And Braindonor, who added a reply while I was typing)I submitted this question quite a long time ago - and I found that I had to spend time to research some of the options again. :w00t:BTW, maybe it's a bit early to say anything about the percentages, but at this point I am mainly surprised at:* 28% wrong answers on the "bitmap on computed" option - I expected this to be lower* 17% wrong answers on the "nonclustered on a view" option - here, I expected a higher failure rateThe other percentages are (so far) about on par with my expectation.[/quote]Good question, Hugo! I guess you could say it was so good, even you learned from it! :-PI'm most surprised that (at the time I answered) 49% thought "clustered with included columns" was a valid answer. Since the clustered index by design includes all columns at the leaf level, the only way I can see choosing that answer is if they were blindly guessing, or they reasoned that it was a trick choice and "with included columns" meant the implicit inclusion of columns rather than the explicit definition of included columns.

RE: Index types 2

Thomas Abraham — Thu, 30 Aug 2012 05:53:32 GMT

Thanks for the question Hugo. I should have stretched out before running that hard early in the morning. I think I strained something! :-) I certainly learned something.

RE: Index types 2

Divine Flame — Thu, 30 Aug 2012 04:17:42 GMT

Very good question! Thanks.

RE: Index types 2

Hugo Kornelis — Thu, 30 Aug 2012 04:03:34 GMT

[quote][b]DugyC (8/30/2012)[/b][hr]Had to answer by process of elimination[/quote]Thanks, Dugy!Frankly, I am a bit disappointed that the site moderators included the number of correct answers. I had included a note that there are multiple correct answers without specifying how many, to make it a true test of knowledge - but apprently, including the number of correct answers for questions with more than one answer is now standard practice, applied to all those questions. :(

RE: Index types 2

DugyC — Thu, 30 Aug 2012 03:54:18 GMT

Thanks for the great question Hugo, got the brain matter working. Had to answer by process of elimination and learned something new on the way. Excellent.Cheers :-D

RE: Index types 2

Hugo Kornelis — Thu, 30 Aug 2012 03:17:31 GMT

[quote][b]Gazareth (8/30/2012)[/b][hr]Nice question Hugo, thanks.Got 3 of the 4 - knew bitmap indexes were available on other DB software, thought it might have been introduced with 2012.Thought 260 columns for columnstore indexes was a slightly random number so didn't go for that :-)[/quote]Thanks, Gazareth!The 260 number is deliberately chosen to exceed 256, 2[sup]8[/sup] or the number of values in a tinyint. This vlaue is a common boundary value in software. (For example, before SQL Server 2008, nonclustered indexes on a table were identified by a tinyint column with two reserved values, leaving room for 254 other values; the actual maximum number was 250 - I think they simply rounded down to a pretty number). So I picked the number 260 to distinguish those who believe in a maximum of 256 columns per columnstore index from those who believe in a larger or no maximum.

RE: Index types 2

Gazareth — Thu, 30 Aug 2012 02:57:25 GMT

Nice question Hugo, thanks.Got 3 of the 4 - knew bitmap indexes were available on other DB software, thought it might have been introduced with 2012.Thought 260 columns for columnstore indexes was a slightly random number so didn't go for that :-)

RE: Index types 2

Hugo Kornelis — Thu, 30 Aug 2012 01:22:01 GMT

Thanks, Koen! (EDIT: And Braindonor, who added a reply while I was typing)I submitted this question quite a long time ago - and I found that I had to spend time to research some of the options again. :w00t:BTW, maybe it's a bit early to say anything about the percentages, but at this point I am mainly surprised at:* 28% wrong answers on the "bitmap on computed" option - I expected this to be lower* 17% wrong answers on the "nonclustered on a view" option - here, I expected a higher failure rateThe other percentages are (so far) about on par with my expectation.

RE: Index types 2

BrainDonor — Thu, 30 Aug 2012 01:20:38 GMT

Me too - quite a bit of hunting to get the right answers. Thanks for the interesting question.

RE: Index types 2

Koen Verbeeck — Thu, 30 Aug 2012 00:16:24 GMT

Great question Hugo, made me do some research :-)

Index types 2

Hugo Kornelis — Wed, 29 Aug 2012 23:22:41 GMT

Comments posted to this topic are about the item [B]Index types 2[/B]