Understanding Indexes

Question

Understanding Indexes

Viewing 12 posts - 16 through 26 (of 26 total)

You must be logged in to reply to this topic. Login to reply

Kiara SSCarpal Tunnel Points: 4185 More actions · Answer 1

Jeff Moden (2/15/2011)
Or... perhaps I'm just being a bit too critical of an introductory article.

I don't think you are.

I've seen plenty of "introduction to indexing" articles that are far easier to read and introduce the concepts behind indexes in a more approachable manner. If this was intended as "Index Theory 101", it seems to include some good basic information, but presented in a way that makes it seem far more complicated than it actually is - particularly when we read the caveats about what is out of scope and what is being deferred to other discussions.

If it's intended as "Practical Index Basics 101 for SQL Server", well, it appears to also miss the mark for its intended audience.

I guess that's why I'm so curious about the original reason this article was written and who the target audience was.

I'd rather refer someone new to the topic to Gail's series that starts here: http://www.sqlservercentral.com/articles/Indexing/68439/.

Ah, well. I've probably devoted too much time today to puzzling over this particular article.

-Ki

Charles Kincaid SSChampion Points: 13593 More actions · Answer 2

First I think that the intent, target, and scope are great. Yes, there is room for revision.

Nice effort, Ron. Everybody gets the blowtorch to the backside for everything posted here. I remember having to suck it up and revise. Welcome to the team.

ATBCharles Kincaid

zebulonpi SSC Journeyman Points: 75 More actions · Answer 3

I think the criticism is less "taking a blowtorch to his backside" and more commenting about how, for an article that clearly states its intended audience is "the junior database administrator (DBA) or others with little or no experience in the design, administration, or optimization of database systems", some of the errors in the article may be taken as correct facts.

A lot of these articles are "fire and forget", in terms of people reading them, so someone new to SQL and indexing may well read this, say "OK, I understand now", and never come back to see any revisions, hence what I see as a concern.

Also, when publishing something intended as educational, AND being reviewed by your peers, you WILL get criticism. Take it as constructive, learn from it, and become better.

Charles Kincaid SSChampion Points: 13593 More actions · Answer 4

zebulonpi (2/15/2011)
Also, when publishing something intended as educational, AND being reviewed by your peers, you WILL get criticism. Take it as constructive, learn from it, and become better.

Which was exactly my point. If there was no criticism then either it was not read or nobody cared as human effort is not perfect. Still the "blowtorch to the backside" was about the cleanest way that I could think of to state how this feels.

ATBCharles Kincaid

roger_os SSC Eights! Points: 831 More actions · Answer 5

I enjoy articles like this because I learn so much from the responses!

Gail Shaw SSC Guru Points: 1004504 More actions · Answer 6

Jeff Moden (2/15/2011)
Since I'm not an expert on indexes that you good folks are, can you confirm that the bolded statements in the above quote are, depending on whether it's a covering index or not, incorrect... or not? I realize that data is stored in the index but that data is actually used if it's a covering index.

A nonclustered index does not just contain an address of the actual row. It contains the index key columns, any include columns defined and either the clustered index key or the RID, depending whether the base is a heap or a cluster. The clustered index key/RID can be considered the 'address' of the row.

Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability

We walk in the dark places no others will enter
We stand on the bridge and no one may pass

zebulonpi SSC Journeyman Points: 75 More actions · Answer 7

Charles Kincaid (2/15/2011)Still the "blowtorch to the backside" was about the cleanest way that I could think of to state how this feels.

As far as FEELING goes, I'll give you that one. 😀

Still, one of the fastest ways to learn is to cover yourself with those heat-resistant tiles from the Space Shuttle and walk blithely into something like this. If you can stand the heat in the kitchen, you learn to cook VERY quickly.

Peter Trast SSCarpal Tunnel Points: 4392 More actions · Answer 8

zebulonpi (2/15/2011)
Charles Kincaid (2/15/2011)Still the "blowtorch to the backside" was about the cleanest way that I could think of to state how this feels.
As far as FEELING goes, I'll give you that one. 😀
Still, one of the fastest ways to learn is to cover yourself with those heat-resistant tiles from the Space Shuttle and walk blithely into something like this. If you can stand the heat in the kitchen, you learn to cook VERY quickly.

Sometimes I am even nervous about participating in the dialog 🙂

Peter Trast
Microsoft Certified ...(insert many literal strings here)
Microsoft Design Architect with Alexander Open Systems

Jeff Moden SSC Guru Points: 1004378 More actions · Answer 9

Grant Fritchey (2/15/2011)
Jeff Moden (2/15/2011)
Grant Fritchey (2/15/2011)
Define a clustered index on the column that appears most frequently in the WHERE clause of SELECT
statements.
But not in JOIN criteria? No, no way. The basic rule for clusters should be the most frequently used access path to the data. This may be primary keys, this may be foreign keys, or it might be simply search criteria, but I wouldn't suggest limiting it to WHERE clauses.
Again, I'm not the "index Ninja" here, but I've also found that a table that suffers huge numbers of inserts can make very good use of a Clustered Index on an auto-numbering column such as an IDENTITY column or, perhaps, a date column to keep page splits to a reasonable level whether or not that column is the most frequently used in WHERE or JOIN criteria.
Have any of you good folks experienced the same or is there some other practice that folks use on a high insertion rate table?
I'm not an "index Niinja" either. I leave that to Gail, but in my experience and with my understanding, you're 100% right on here. Again, I like my own definition of "most frequently used access path" because that can apply to inserts as well as updates, deletes or selects. It doesn't matter. Since the cluster holds the data, you need to take advantage of that fact in whatever way is most advantageous to the system you're working on.

Thanks, Grant.

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)

zebulonpi SSC Journeyman Points: 75 More actions · Answer 10

Peter Trast (2/15/2011)Sometimes I am even nervous about participating in the dialog 🙂

Try physically working with quite a few knowledgable SQL people, using a 'not like' against a 650 million record table in your 11 table joined query, and then asking said knowledgable people why the server was so slow today...

At least in the forums they can't physically choke you... 😀

Jeff Moden SSC Guru Points: 1004378 More actions · Answer 11

GilaMonster (2/15/2011)
Jeff Moden (2/15/2011)
Since I'm not an expert on indexes that you good folks are, can you confirm that the bolded statements in the above quote are, depending on whether it's a covering index or not, incorrect... or not? I realize that data is stored in the index but that data is actually used if it's a covering index.
A nonclustered index does not just contain an address of the actual row. It contains the index key columns, any include columns defined and either the clustered index key or the RID, depending whether the base is a heap or a cluster. The clustered index key/RID can be considered the 'address' of the row.

Thanks, Gail. I guess it was just the way it was worded in the article. I thought I knew better (according to the possibly incorrect interpretation on my part) but needed to confirm it with someone that knew for sure. As usual, you difinitely came through on the clarification. 🙂

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)

samuel_david_piper Grasshopper Points: 19 More actions · Answer 12

There are two points in the article that I wanted to respond to:

Avoid creating an index in response to the poor performance of a single query.

This needs to be fleshed out more - indexes are created primarily in response to poor query performance. The question is, which queries should be indexed? If a given query is ad hoc and never used by a calling application then there is no immediate benefit to indexing it, regardless of its performance. If the same query was then called hundreds of times a minute by different users then we should create indexes to improve the performance of that query.

Index optimization is a complex topic, and ultimately I don't think it can be reduced to blanket statements such as this one.

When queries against a table do not include a WHERE clause there is no benefit to using a
nonclustered index.

That's incorrect. For example, JOIN or ORDER BY conditions can benefit from a nonclustered index. Imagine a table of contacts where the most frequent query was:

SELECT ContactID, FirstName, LastName FROM Contact ORDER BY LastName ASC, FirstName ASC

If you only have a clustered index on the IDENTITY column then SQL Server will have to scan and sort the results for these two columns every time the query is executed. However if you define a nonclustered index on LastName ASC, FirstName ASC, then SQL Server will be able to read just that index to satisfy the entire query, without needing to do any sorting as this is already done whenever the index is updated.

This will scale particularly well as the table size increases as there is no increasing cost for sorting as there is when only a clustered index is available. There is also now support for seeking values in these columns more efficiently as well.

As others have said, any kind of data access path can potentially benefit from a nonclustered index, and the best way to know what those paths are is to look at the execution plan.

Also for a beginner article, there really needs to be a discussion of how to actually create an index in the first place, either via Management Studio or T-SQL.