Lately, I have been doing some reading on indexes of all types and the main advice is to put the clustered index on the primary key of the table, but what if the primary key actually is not used in a query (via a select or join) and is just put for purely relational purposes, so in this case it is not queried against. Example, say I have a car_parts table and it contains 3 columns, car_part_id, car_part_no, and car_part_title. car_part_id is the unique primary key identity column. In this case car_part_no is unique as well and is most likely car_part_title. car_part_no is what is most queried against, so doesn't it make sense to put the clustered index on that column instead of car_part_id? The basics of the question is what column should actually have the clustered index since you are only allowed one of them?
An index, clustered or non clustred, can be used by the query optimizer if and only if the leftmost key in the index is filtered on. So if you define an index on columns (A, B, C), a WHERE condition on B=#b, on C=#c or on B=#b AND C=#c will not fully leverage the index (see note). This applies also to join conditions. Any WHERE filter that includes A will consider the index: A=#a or A=#a AND B=#b or A=#a AND C=#c or A=#a AND B=#b AND C=#c.
So in your example if you make the clustred index on part_no as the leftmost key, then a query looking for a specific part_id will not use the index and a separate non-clustered index must exist on part-id.
Now about the question which of the many indexes should be the clustered one. If you have several query patterns that are about the same importance and frequency and contradict each other on terms of the keys needed (eg. frequent queries by either part_no or part_id) then you take other factors into consideration:
width: the clustered index key is used as the lookup key by all other non-clustered indexes. So if you choose a wide key (say two uniquidentifier columns) then you are making all the other indexes wider, thus consuming more space, generating more IO and slowing down everything. So between equaly good keys from a read point of view, choose the narrowest one as clustered and make the wider ones non-clustered.
contention: if you have specific patterns of insert and delete try to separate them physically so they occur on different portions of the clustered index. Eg. if the table acts as a queue with all inserts at one logical end and all deletes at the other logical end, try to layout the clustered index so that the physical order matches this logical order (eg. enqueue order).
partitioning: if the table is very large and you plan to deploy partioning then the partitioning key must be the clustered index. Typical example is historical data that is archived using a sliding window partitioning scheme. Even thow the entities have a logical primary key like 'entity_id', the clustred index is done by a datetime column that is also used for the partitioning function.
stability: a key that changes often is a poor candidate for a clustered key as each update the clustered key value and force all non-clustered indexes to update the lookup key they store. As an update of a clustered key will also likely relocate the record into a different page it can cause fragmentation on the clustered index.
Note: not fully leverage as sometimes the engine will choose an non-clustered index to scan instead of the clustered index simply because is narrower and thus has fewer pages to scan. In my example if you have an index on (A, B, C) and a WHERE filter on B=#b and the query projects C, the index will be likely used but not as a seek, as a scan, because is still faster than a full clustered scan (fewer pages).
Kimberly Tripp is always one of the best sources on insights on indexing.
See her blog post "Ever-increasing clustering key - the Clustered Index Debate - again!" in which she quite clearly lists and explains the main requirements for a good clustering key - it needs to be:
Unique
Narrow
Static
and best of all, if you can manage:
ever-increasing
Taking all this into account, an INT IDENTITY (or BIGINT IDENTITY if you really need more than 2 billion rows) works out to be the best choice in the vast majority of cases.
One thing a lot of people don't realize (and thus don't take into account when making their choice) is the fact that the clustering key (all the columns that make up the clustered index) will be added to each and every index entry for each and every non-clustered index on your table - thus the "narrow" requirement becomes extra important!
Also, since the clustering key is used for bookmark lookups (looking up the actual data row when a row is found in a non-clustered index), the "unique" requirement also becomes very important. So important in fact, that if you choose a (set of) column(s) that is/are not guaranteed to be unique, SQL Server will add a 4-byte uniquefier to each row --> thus making each and every of your clustered index keys extra wide ; definitely NOT a good thing.
Marc
Clustered indexes are good when you query ranges of data. For example
SELECT * FROM theTable WHERE age BETWEEN 10 AND 20
The clustered index arranges rows in the particular order on your computer disk. That's why rows with age = 10 will be next to each other, and after them there will be rows with age = 11, etc.
If you have exact select, like this:
SELECT * FROM theTable WHERE age = 20
the non-clustered index is also good. It doesn't rearrange data on your computer disk, but it builds special tree with a pointers to the rows you need.
So it strongly depends on the type of queries you perform.
Keep in mind the usage patterns; If you are almost always querying the DB on the car_part_no, then it would probably be beneficial for it to be clustered on that column.
However, don't forget about joins; If you are most often joining to the table and the join uses the car_part_id field, then you have a good reason to keep the cluster on car_part_id.
Something else to keep in mind (less so in this case, but generally when considering clustered indexes) is that the clustered index will appear implicitly in every other index on the table; So for example, if you were to index car_part_title, that index will also include the car_part_id implicitly. This can affect whether or not an index covers a query and also affects how much disk space the index will take (which affects memory usage, etc).
The clustered index should go on the column that will be the most queried. This includes joins, as a join must access the table just like a direct query, and find the rows indicated.
You can always rebuild your indexes later on if your application changes and you find you need to optimize a table with a different index structure.
Some additional guidelines for deciding on what to cluster your table on can be found on MSDN here: Clustered Index Design Guidelines.
Related
This question already has answers here:
Better to use a Clustered index or a Non-Clustered index with included columns?
(3 answers)
Closed 9 years ago.
I have a table with columns a,b,c,d,e,f,g that has roughly 500,000 rows.
There is a query that gets run very often that does a SELECT * FROM table WHERE a = #a AND b = #b AND c = #c.
Is it better to create a clustered index on a, b, and c, OR am I better off creating a non-clustered index on a, b, and c INCLUDE (d, e, f, g).
Not sure the include would help speed up the query since the select * was issued.
Any help would be appreciated!
A clustered index would be the fastest for that SELECT, but it may not necessarily be correct choice.
A clustered index determines the order in which records are physically stored (which is why you can only have one per table). So while it would be the fastest for THAT query, it may slow down other queries and could KILL updates and inserts if one of those columns was changing, which could mean that the record would need to be physically re-located.
An INCLUDE would also speed up that query at the expense of extra storage and extra index maintenance if any of those fields (including the included fields) were updated.
I would START with a non-clustered index on a, b, and c and see if that gets your performance to a reasonable level. Anything more could just be trading speed in one area for slowness in another.
The clustered index will be faster.
With SELECT *, both your clustered and non-clustered (with include-all) contain all the columns within each page. However, the non-clustered index ALSO contains a reference back to the clustered key - this is required in case you add more columns to the table, but really also because all indexes (except indexed views) are pointers to the data pages. The NCI will not feature the new columns (fixed include list) but the data pages will.
SQL Server may be smart enough to find out that SELECT * can be fulfilled by an INDEX SCAN on the NCI (+includes) without a bookmark lookup back to the data pages, but even then, that index scan will be one column wider than the equivalent clustered index scan.
It is normally not a good idea to have a 3-column clustering key. You may consider an alternative of using a simple single-column identity clustering key, and creating an indexed view clustered around the 3 columns.
The answer to the question as stated in your subject line in general is no. Because you generally would much prefer to have the narrowest covering (probably non-clustered) index.
But in your case you are selecting * so if the clustered index is good enough match to your seek criteria it's always going to be picked, since anything narrower will need to do a bookmark lookup.
So this raises a big question of why this query is the way it is, whether there is a better choice of clustered index in general for your app (narrow, static, increasing, unique), and whether you really need to be getting all the columns. Because neither of the two options you give is really that typical of a good design.
500000 rows is fairly small, but if performance is an issue, you want to see how many rows are fitting per page and whether you could improve that be being more selective in your query and having a covering non-clustered index.
Your Clustered index is the order the data is stored in the table, so you can only have one clustered index per table. If you create a new index (non-clustered by default) sure the columns are defined in the index in the same order they are used in the WHERE clause that will allow SQL do a direct index scan to find the record(s) you're looking for.
I have a table with two very important fields:
id INT identity(1,1) PRIMARY KEY
identifiersortcode VARCHAR(900)
My app always sorts and pages search results in the UI based on identifiersortcode, but all table joins (and they are legion) are on the id field. (Aside: yes, the sort code really is that long. There's a strong BL reason.)
Also, due to O/RM use, most SELECT statements are going to pull almost every column.
Currently, the clustered index is on id, but I'm wondering if the TOP / ORDER BY portion of most queries would make identifiersortcode a more attractive option as the clustered key, even considering all of the table joins going on.
Inserts on the table and changes to the identifiersortcode are limited enough that changing my clustered index would be a problem for insert/update operations.
Trying to make the sort code's non-clustered index a covering index (using INCLUDE) is not a good option. There are a number of large columns, and some of them have a lot of update activity.
Kimberly L. Tripp's criteria for a clustered index are that it be:
Unique
Narrow
Static
Ever Increasing
Based on that, I'd stick with your integer identity id column, which satisfies all of the above. Your identifiersortcode would fail most, if not all, of those requirements.
To correctly determine which field will benefit most from the clustered index, you need to do some homework. The first thing that you should consider is the selectivity of your joins. If your execution plans filter rows from this table FIRST, then join on the other tables, then you are not really benefiting from having the clustered index on the primary key, and it makes more sense to have it on the sort key.
If however, your joins are selective on other tables (they are filtered, then an index seek is performed to select rows from this table), then you need to compare the performance of the change manually versus the status quo.
Currently, the clustered index is on id, but I'm wondering if the TOP / ORDER BY portion of most queries would make identifiersortcode a more attractive option as the clustered key, even considering all of the table joins going on.
Making identifiersortcode a CLUSTERED KEY will only help if it is used both in filtering and ordering conditions.
This means that it is chosen a leading table in all your joins and uses Clustered Index Scan or Clustered Index Range Scan access path.
Otherwise, it will only make the things worse: first, all secondary indexes will be larger in size; second, inserts in non-increasing order will result in page splits which will make them run longer and result in a larger table.
Why, for God's sake, does your identifier sort code need to be 900 characters long? If you really need 900 characters to be distinct for sorting, it should probably be broken up into multiple fields.
Appart from repeating what Chris B. said, I think you should really stick to your current PK, since - as you said - all joins are on the Id.
I guess you already have indexed the identifiersortcode....
Nevertheless, IF you have performance issues, would reaaly think twice about this ##"%$£ identifiersortcode !-)
What happens if a clustered index is not unique? Can it lead to bad performance because inserted rows flow to an "overflow" page of some sorts?
Is it "made" unique and if so how? What is the best way to make it unique?
I am asking because I am currently using a clustered index to divide my table in logical parts, but the performance is so-so, and recently I got the advice to make my clustered indexes unique. I'd like a second opinion on that.
They don't have to be unique but it certainly is encouraged.
I haven't encountered a scenario yet where I wanted to create a CI on a non-unique column.
What happens if you create a CI on a non-unique column
If the clustered index is not a unique
index, SQL Server makes any duplicate
keys unique by adding an internally
generated value called a uniqueifier
Does this lead to bad performance?
Adding a uniqueifier certainly adds some overhead in calculating and in storing it.
If this overhead will be noticable depends on several factors.
How much data the table contains.
What is the rate of inserts.
How often is the CI used in a select (when no covering indexes exist, pretty much always).
Edit
as been pointed out by Remus in comments, there do exist use cases where creating a non-unique CI would be a reasonable choice. Me not having encountered one off those scenarios merely shows my own lack of exposure or competence (pick your choice).
I like to check out what The Queen of Indexing, Kimberly Tripp, has to say on the topic:
I'm going to start with my recommendation for the Clustering Key - for a couple of reasons. First, it's an easy decision to make and second, making this decision early helps to proactively prevent some types of fragmentation. If you can prevent certain types of base-table fragmentation then you can minimize some maintenance activities (some of which, in SQL Server 2000 AND less of which, in SQL Server 2005) require that your table be offline. OK, I'll get to the rebuild stuff later.....
Let's start with the key things that I look for in a clustering key:
* Unique
* Narrow
* Static
Why Unique?
A clustering key should be unique because a clustering key (when one exists) is used as the lookup key from all non-clustered indexes. Take for example an index in the back of a book - if you need to find the data that an index entry points to - that entry (the index entry) must be unique otherwise, which index entry would be the one you're looking for? So, when you create the clustered index - it must be unique. But, SQL Server doesn't require that your clustering key is created on a unique column. You can create it on any column(s) you'd like. Internally, if the clustering key is not unique then SQL Server will “uniquify” it by adding a 4-byte integer to the data. So if the clustered index is created on something which is not unique then not only is there additional overhead at index creation, there's wasted disk space, additional costs on INSERTs and UPDATEs, and in SQL Server 2000, there's an added cost on a clustereD index rebuild (which because of the poor choice for the clustering key is now more likely).
Source: Ever-increasing clustering key debate - again!
Do clustered indexes have to be unique?
They don't, and there are times where it's better if they're not.
Consider a table with a semi-random, unique EmployeeId, and a DepartmentId for each employee: if your select statement is
SELECT * FROM EmployeeTable WHERE DepartmentId=%DepartmentValue%
then it's best for performance if the DepartmentId is the clustered index even though (or even especially because) it's not the unique index (best for performance because it ensures that all the records within a given DepartmentId are clustered).
Do you have any references?
There's Clustered Index Design Guidelines for example, which says,
With few exceptions, every table
should have a clustered index defined
on the column, or columns, that offer
the following:
Can be used for frequently used queries.
Provide a high degree of uniqueness.
Can be used in range queries.
My understanding of "high degree of uniqueness" for example is that it isn't good to choose "Country" as the clusted index if most of your queries want to select the records within a given town.
If you are tuning an old DB this is a Godsend. I am working on Perf issues on a 20-year-old DB. It has nonclustered PKs with 3 - 8 columns. Instead of using all 8 columns to be unique I can pick one column with broad distribution, and it applies a Uniqueifier. It is an Int but by using a column like Project ID it can handle 2147483647 unique projectIDs which is enough for most use-cases. If it is not enough add a second or third column to the cluster.
This works without any coding modification in the App layer. 20 years in production and management doesn't have to order a major rewrite.
In many places it's recommended that clustered indexes are better utilized when used to select range of rows using BETWEEN statement. When I select joining by foreign key field in such a way that this clustered index is used, I guess, that clusterization should help too because range of rows is being selected even though they all have same clustered key value and BETWEEN is not used.
Considering that I care only about that one select with join and nothing else, am I wrong with my guess ?
Discussing this type of issue in the absolute isn't very useful.
It is always a case-by-case situation !
Essentially, access by way of a clustered index saves one indirection, period.
Assuming the key used in the JOIN, is that of the clustered index, in a single read [whether from an index seek or from a scan or partial scan, doesn't matter], you get the whole row (record).
One problem with clustered indexes, is that you only get one per table. Therefore you need to use it wisely. Indeed in some cases, it is even wiser not to use any clustered index at all because of INSERT overhead and fragmentation (depending on the key and the order of new keys etc.)
Sometimes one gets the equivalent benefits of a clustered index, with a covering index, i.e. a index with the desired key(s) sequence, followed by the column values we are interested in. Just like a clustered index, a covering index doesn't require the indirection to the underlying table. Indeed the covering index may be slightly more efficient than the clustered index, because it is smaller.
However, and also, just like clustered indexes, and aside from the storage overhead, there is a performance cost associated with any extra index, during INSERT (and DELETE or UPDATE) queries.
And, yes, as indicated in other answers, the "foreign-key-ness" of the key used for the clustered index, has absolutely no bearing on the the performance of the index. FKs are constraints aimed at easing the maintenance of the integrity of the database but the underlying fields (columns) are otherwise just like any other field in the table.
To make wise decisions about index structure, one needs
to understands the way the various index types (and the heap) work
(and, BTW, this varies somewhat between SQL implementations)
to have a good image of the statistical profile of the database(s) at hand:
which are the big tables, which are the relations, what's the average/maximum cardinality of relation, what's the typical growth rate of the database etc.
to have good insight regarding the way the database(s) is (are) going to be be used/queried
Then and only then, can one can make educated guesses about the interest [or lack thereof] to have a given clustered index.
I would ask something else: would it be wise to put my clustered index on a foreign key column just to speed a single JOIN up? It probably helps, but..... at a price!
A clustered index makes a table faster, for every operation. YES! It does. See Kim Tripp's excellent The Clustered Index Debate continues for background info. She also mentions her main criteria for a clustered index:
narrow
static (never changes)
unique
if ever possible: ever increasing
INT IDENTITY fulfills this perfectly - GUID's do not. See GUID's as Primary Key for extensive background info.
Why narrow? Because the clustering key is added to each and every index page of each and every non-clustered index on the same table (in order to be able to actually look up the data row, if needed). You don't want to have VARCHAR(200) in your clustering key....
Why unique?? See above - the clustering key is the item and mechanism that SQL Server uses to uniquely find a data row. It has to be unique. If you pick a non-unique clustering key, SQL Server itself will add a 4-byte uniqueifier to your keys. Be careful of that!
So those are my criteria - put your clustering key on a narrow, stable, unique, hopefully ever-increasing column. If your foreign key column matches those - perfect!
However, I would not under any circumstances put my clustering key on a wide or even compound foreign key. Remember: the value(s) of the clustering key are being added to each and every non-clustered index entry on that table! If you have 10 non-clustered indices, 100'000 rows in your table - that's one million entries. It makes a huge difference whether that's a 4-byte integer, or a 200-byte VARCHAR - HUGE. And not just on disk - in server memory as well. Think very very carefully about what to make your clustered index!
SQL Server might need to add a uniquifier - making things even worse. If the values will ever change, SQL Server would have to do a lot of bookkeeping and updating all over the place.
So in short:
putting an index on your foreign keys is definitely a great idea - do it all the time!
I would be very very careful about making that a clustered index. First of all, you only get one clustered index, so which FK relationship are you going to pick? And don't put the clustering key on a wide and constantly changing column
An index on the FK column will help the JOIN because the index itself is ordered: clustered just means that the data on disk (leaf) is ordered rather then the B-tree.
If you change it to a covering index, then clustered vs non-clustered is irrelevant. What's important is to have a useful index.
It depends on the database implementation.
For SQL Server, a clustered index is a data structure where the data is stored as pages and there are B-Trees and are stored as a separate data structure. The reason you get fast performance, is that you can get to the start of the chain quickly and ranges are an easy linked list to follow.
Non-Clustered indexes is a data structure that contains pointers to the actual records and as such different concerns.
Refer to the documentation regarding Clustered Index Structures.
An index will not help in relation to a Foreign Key relationship, but it will help due to the concept of "covered" index. If your WHERE clause contains a constraint based upon the index. it will be able to generate the returned data set faster. That is where the performance comes from.
The performance gains usually come if you are selecting data sequentially within the cluster. Also, it depends entirely on the size of the table (data) and the conditions in your between statement.
I have inherited a database where there are clustered indexes and additional duplicate indexes for each of the clustered index.
i.e
IX_PrimaryKey is a clustered index on the column ID.
IX_ID is a non clustered index on the column ID.
I want to clean up these duplicate non clustered indexes and I wanted to check to see if anyone could think of a reason to do this.
Can anyone think of a performance benefit for doing this?
For exact same indexes, there's no performance gain. Actually, it incurs performance loss in insertion and updates. However, if there are multicolumn indexes with different column order, there might be a valid reason for them.
Maybe I'm not thinking hard enough, but I can't see any reason to do this; the nature of the clustered index is that the data is organized in the order of the index. It seems that the extra index is a complete waste.
Digging through BOL and watching this question, though ...
There seems no sensible reason for doing this, and there is a performance hit.
The only thing I could think of to do this is to create an index with an incredibly narrow row width so that the rows per page was very high, making it very quick to scan / seek. But since it contains no other fields (except the clustered key, which is the same value) I still cannot see a reason for it.
It's quite possible the original creator was not aware that the PK was defaulting to a clustered index and created an NC index without realising it was a duplicate.
I presume what would have happened is that SQL Server would have automatically created clustered index when a primary key constraint was specified (this would happen if another index (non-clustered/clustered) is not present already) and then some one might have created a non-clustered index for the primary key column.
Such a scenario would:
Have some adverse effect on performance as indexes are updated when inserts/deletes/updates happen.
Use additional disk space.
Might lead to deadlocks.
Would contribute to more time in backup/restore of database.
cheers
It will be a waste to create a clustered primary key. Unless you have query that search for records using WHERE ID = 10 ?
You may want to create a clustered index on the column which will be frequently queried on WHERE City = 'Sydney'. Clustered means that SQL will group the data in the table based on the clustered index. By grouping the City values in the table means SQL can search for data quicker.
Storing two indexes over the same data is a waste of disk space and the processing needed to maintain the data.
However, I can imagine a product which depends on the existence of an index named IX_PrimaryKey. E.G.
string queryPattern = "select * from {0} as t with (index(IX_PrimaryKey))";
You can make the argument that the clustered index itself occupies much less space than the others, since the leaf is the actual data. On the other hand, the clustered index can be more susceptible to page splitting, and some indexes are better non-clustered.
Putting this together, I can definitely think of scenarios where removing the duplicate indexes would be a Bad Thing:
Code like above which depends on a known index name.
Code which can alter the clustered index to any of the non-clustered indexes.
Code which uses the presence/absence of IX_PrimaryKey to treat the table in a certain way.
I don't consider any of these good design, but I can definitely imagine someone doing it. (Have you posted this to DailyWTF?)
There are cases where it makes sense to have overlapping indexes which are not identical:
create index IX_1 on table1 (ID)
create index IX_2 on table1 (ID, TYPE, ORDER_DATE, TOTAL_CHARGES)
If you are looking up strictly by ID, SQL can optimize and use IX_1. If you are running a query based on ID, TYPE, ORDER_DATE and summing up TOTAL_CHARGES, SQL can use IX_2 as a "covering index", satisfying all the query details from the index without ever touching the table. Generally this is something you add in the course of performance tuning, after extensive testing.
Looking at your given example of two indexes on exactly the same field, I don't see a great fit. Perhaps SQL can use IX_ID as a "covering index" when checking for the existence of a value and bypass some blocking on IX_PrimaryKey?