Sql-server – What happens when you change a heap into a clustered table in SQL Server

clustered-indexheapsql server

As I understand, a heap is an unordered object. When you insert a record, SQL Server use IAM page to get the pages that belong to this heap, and use PFS page to find a particular one which has enough space to accommodate this record and insert into it.

When you create a clustered index on it, it becomes a clustered table and the clustered index itself becomes the table. But as the clustered index and the original heap are two different structures, does SQL Server create a new structure (the clustered index) and moves everything from the help to the new structure and then drops the heap?

There are a lot of stuffs that can be defined on a table, like triggers, constrains, permissions etc. If my assumption is true, that means SQL Server also moves all these stuff to the new structure. I didn't find any related information in the documentation. Is my understanding correct?

Best Answer

Yes, when you create a clustered index on a heap the rows are all sorted moved to the new clustered index. Any non-clustered indexes are rebuilt with the new clustered index key as the row locator.

It's the same table, though, so triggers, constraints, etc don't have to change.

The reverse is not true, however. When you drop a clustered index on a table, the leaf-level pages of the clustered index are left in-place and they become the new heap. Nonclustered indexes are still rebuilt as the row locator switches from the CI key to the rowid (file:page:slot).

SELECT performance

To check performanc numbers I performed the following queries once on the heap table and once on the clust table:

select * from heap/clust where group between 5678910 and 5679410
select * from heap/clust where group between 6234567 and 6234967
select * from heap/clust where group between 6455429 and 6455729
select * from heap/clust where group between 6655429 and 6655729
select * from heap/clust where group between 6955429 and 6955729
select * from heap/clust where group between 7195542 and 7155729

The results of this benchmark are for the heap:

rows  reads CPU   Elapsed 
----- ----- ----- --------
1503  1510  31ms  309ms
401   405   15ms  283ms
2700  2709  0ms   472ms
0     3     0ms   30ms
2953  2962  32ms  257ms
0     0     0ms   0ms

Update on 9 Mar 2011: cmd.CommandText = "select * from heap where group between @id and @id+1000";

721 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1001      69788    6368         -         
Cpu            15        374      37   0.00754
Reads        1069      91459    7682   1.20155
Writes          0          0       0   0.00000
Duration   0.3716   282.4850 10.3672   0.00180

End of Update on 9 Mar 2011.

for the table clust the results are:

rows  reads CPU   Elapsed 
----- ----- ----- --------
1503  4827  31ms  327ms
401   1241  0ms   242ms
2700  8372  0ms   410ms
0     3     0ms   0ms
2953  9060  47ms  213ms
0     0     0ms   0ms

Update on 9 Mar 2011: cmd.CommandText = "select * from clust where group between @id and @id+1000";

721 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1001      69788    6056         -
Cpu            15        468      38   0.00782
Reads        3194     227018   20457   3.37618
Writes          0          0       0       0.0
Duration   0.3949   159.6223 11.5699   0.00214

End of Update on 9 Mar 2011.

SELECT WITH JOIN performance

cmd.CommandText = "select * from heap/clust h join keys k on h.group = k.group where h.group between @id and @id+1000";

The results of this benchmark are for the heap:

873 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1009       4170    1683         -
Cpu            15         47      18   0.01175
Reads        2145       5518    2867   1.79246
Writes          0          0       0   0.00000
Duration   0.8215   131.9583  1.9095   0.00123

The results of this benchmark are for the clust:

865 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1000       4143    1685         -
Cpu            15         47      18   0.01193
Reads        5320      18690    8237   4.97813
Writes          0          0       0   0.00000
Duration   0.9699    20.3217  1.7934   0.00109

UPDATE performance

The second batch of queries are update statements:

update heap/clust set amount = amount + 0 where group between 5678910 and 5679410
update heap/clust set amount = amount + 0 where group between 6234567 and 6234967
update heap/clust set amount = amount + 0 where group between 6455429 and 6455729
update heap/clust set amount = amount + 0 where group between 6655429 and 6655729
update heap/clust set amount = amount + 0 where group between 6955429 and 6955729
update heap/clust set amount = amount + 0 where group between 7195542 and 7155729

the results of this benchmark for the heap:

rows  reads CPU   Elapsed 
----- ----- ----- -------- 
1503  3013  31ms  175ms
401   806   0ms   22ms
2700  5409  47ms  100ms
0     3     0ms   0ms
2953  5915  31ms  88ms
0     0     0ms   0ms

Update on 9 Mar 2011: cmd.CommandText = "update heap set amount = amount + @id where group between @id and @id+1000";

811 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1001      69788    5598       811         
Cpu            15        873      56   0.01199
Reads        2080     167593   11809   2.11217
Writes          0       1687     121   0.02170
Duration   0.6705   514.5347 17.2041   0.00344

End of Update on 9 Mar 2011.

the results of this benchmark for the clust:

rows  reads CPU   Elapsed 
----- ----- ----- -------- 
1503  9126  16ms  35ms
401   2444  0ms   4ms
2700  16385 31ms  54ms
0     3     0ms   0ms 
2953  17919 31ms  35ms
0     0     0ms   0ms

Update on 9 Mar 2011: cmd.CommandText = "update clust set amount = amount + @id where group between @id and @id+1000";

853 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1001      69788    5420         -
Cpu            15        594      50   0.01073
Reads        6226     432237   33597   6.20450
Writes          0       1730     110   0.01971
Duration   0.9134   193.7685  8.2919   0.00155

End of Update on 9 Mar 2011.

DELETE benchmarks

the third batch of queries I ran are delete statements

delete heap/clust where group between 5678910 and 5679410
delete heap/clust where group between 6234567 and 6234967
delete heap/clust where group between 6455429 and 6455729
delete heap/clust where group between 6655429 and 6655729
delete heap/clust where group between 6955429 and 6955729
delete heap/clust where group between 7195542 and 7155729

The result of this benchmark for the heap:

rows  reads CPU   Elapsed 
----- ----- ----- -------- 
1503  10630 62ms  179ms
401   2838  0ms   26ms
2700  19077 47ms  87ms
0     4     0ms   0ms
2953  20865 62ms  196ms
0     4     0ms   9ms

Update on 9 Mar 2011: cmd.CommandText = "delete heap where group between @id and @id+1000";

724 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts     192      69788    4781         -
Cpu            15        499      45   0.01247
Reads         841     307958   20987   4.37880
Writes          2       1819     127   0.02648
Duration   0.3775  1534.3383 17.2412   0.00349

End of Update on 9 Mar 2011.

the result of this benchmark for the clust:

rows  reads CPU   Elapsed 
----- ----- ----- -------- 
1503  9228  16ms  55ms
401   3681  0ms   50ms
2700  24644 46ms  79ms
0     3     0ms   0ms
2953  26955 47ms  92ms
0     3     0ms   0ms

Update on 9 Mar 2011:

cmd.CommandText = "delete clust where group between @id and @id+1000";

751 Rows have > 0 CPU and affect more than 0 rows

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts     144      69788    4648         -
Cpu            15        764      56   0.01538
Reads         989     458467   30207   6.48490
Writes          2       1830     127   0.02694
Duration   0.2938  2512.1968 24.3714   0.00555

End of Update on 9 Mar 2011.

INSERT benchmarks

The last part of the benchmark is the execution of insert statements.

insert into heap/clust (...) values (...), (...), (...), (...), (...), (...)

The result of this benchmark for the heap:

rows  reads CPU   Elapsed 
----- ----- ----- -------- 
6     38    0ms   31ms

Update on 9 Mar 2011:

string str = @"insert into heap (group, currency, year, period, domain_id, mtdAmount, mtdAmount, ytdAmount, amount, ytd_restated, restated, auditDate, auditUser)
                    values";

                    for (int x = 0; x < 999; x++)
                    {
                        str += string.Format(@"(@id + {0}, 'EUR', 2012, 2, 0, 100, 100, 1000 + @id,1000, 1000,1000, current_timestamp, 'test'),  ", x);
                    }
                    str += string.Format(@"(@id, 'CAD', 2012, 2, 0, 100, 100, 1000 + @id,1000, 1000,1000, current_timestamp, 'test') ", 1000);

                    cmd.CommandText = str;

912 statements have > 0 CPU

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1000       1000    1000         -
Cpu            15       2138      25   0.02500
Reads        5212       7069    6328   6.32837
Writes         16         34      22   0.02222
Duration   1.6336   293.2132  4.4009   0.00440

End of Update on 9 Mar 2011.

The result of this benchmark for the clust:

rows  reads CPU   Elapsed 
----- ----- ----- -------- 
6     50    0ms   18ms

Update on 9 Mar 2011:

string str = @"insert into clust (group, currency, year, period, domain_id, mtdAmount, mtdAmount, ytdAmount, amount, ytd_restated, restated, auditDate, auditUser)
                    values";

                    for (int x = 0; x < 999; x++)
                    {
                        str += string.Format(@"(@id + {0}, 'EUR', 2012, 2, 0, 100, 100, 1000 + @id,1000, 1000,1000, current_timestamp, 'test'),  ", x);
                    }
                    str += string.Format(@"(@id, 'CAD', 2012, 2, 0, 100, 100, 1000 + @id,1000, 1000,1000, current_timestamp, 'test') ", 1000);

                    cmd.CommandText = str;

946 statements have > 0 CPU

Counter   Minimum    Maximum Average  Weighted
--------- ------- ---------- ------- ---------
RowCounts    1000       1000    1000         -      
Cpu            15       2403      21   0.02157
Reads        6810       8997    8412   8.41223
Writes         16         25      19   0.01942
Duration   1.5375   268.2571  6.1463   0.00614

End of Update on 9 Mar 2011.

Conclusions

Although there are more logical reads going on when accessing the table with the clustered & the nonclustered index (while using the nonclustered index) the performance results are:

SELECT statements are comparable
UPDATE statements are faster with a clustered index in place
DELETE statements are faster with a clustered index in place
INSERT statements are faster with a clustered index in place

Of course my benchmark was very limited on a specific kind of table and with a very limited set of queries, but I think that based on this information we can already start saying that it is virtually always better to create a clustered index on your table.

Update on 9 Mar 2011:

As we can see from the added results, the conclusions on the limited tests were not correct in every case.

Weighted Duration

The results now indicate that the only statements which benefit from the clustered index are the update statements. The other statements are about 30% slower on the table with clustered index.

Some additional charts where I plotted the weighted duration per query for heap vs clust. Weighted Duration heap vs clustered for Select

Weighted Duration heap vs clustered for Join

Weighted Duration heap vs clustered for Update

Weighted Duration heap vs clustered for Delete

As you can see the performance profile for the insert statements is quite interesting. The spikes are caused by a few data points which take a lot longer to complete. Weighted Duration heap vs clustered for Insert

End of Update on 9 Mar 2011.

Sql-server – What happens to non clustered index when included column is updated using update statement

The blog post you reference also indicates how you could have answered this yourself.

If you execute

SET SHOWPLAN_TEXT ON;
GO
UPDATE T SET A = 19;

The plan looks like

  |--Clustered Index Update(OBJECT:([AdventureWorks2008].[dbo].[T].[TPK]), OBJECT:([AdventureWorks2008].[dbo].[T].[TF]), SET:([AdventureWorks2008].[dbo].[T].[A] = [@1]))
       |--Compute Scalar(DEFINE:([Expr1009]=[Expr1009]))
            |--Compute Scalar(DEFINE:([Expr1009]=CASE WHEN CASE WHEN [AdventureWorks2008].[dbo].[T].[A] = [@1] THEN (1) ELSE (0) END THEN (0) ELSE (1) END))
                 |--Top(ROWCOUNT est 0)
                      |--Index Scan(OBJECT:([AdventureWorks2008].[dbo].[T].[TF]), ORDERED FORWARD)

showing a per-row / narrow plan where the index TF is listed as one of the objects updated.

Best Answer

Related Solutions

Sql-server – Performance of Non Clustered Indexes on Heaps vs Clustered Indexes

SELECT performance

SELECT WITH JOIN performance

UPDATE performance

DELETE benchmarks

INSERT benchmarks

Conclusions

Sql-server – What happens to non clustered index when included column is updated using update statement

Related Question