Want to protect your cyber security and still get fast solutions? Ask a secure question today.Go Premium

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 459
  • Last Modified:

Use of index hints in MS SQL. Required? How does this work with hibernate?

I have a query like this:

select * from table1,table2 where table2.table1id = table1.id;

table2.table1id is a foreign key that points to a corresponding entry in table1. table1 and table2 have primary keys, both called id.

table2 is also a secondary index, called myindex. Here is my problem. During tests,

Microsoft SQL Server wants to use the primary key of table2 (id) and is scanning the table1 table for the matching id to internally perform the join, instead of using the “table1id” index directly.

I know I can formulate my query using a hint to tell SQL to use the secondary key directly.

Here are my questions:

Is this necessary or are their ways to set up the database indexes so that hints are unnecessary.
This is currently written in straight SQL embedded in my Java program. If I move to hibernate, does hibernate automatically generate the hints for me?
0
efamilant
Asked:
efamilant
1 Solution
 
UnifiedISCommented:
If you use a join statement does it still use the other index?
select * from table1 t1
INNER JOIN table2  t2
ON t2.table1id = t1.id
0
 
PortletPaulCommented:
you definitely should be using ANSI join syntax (like that shown by UnifiedIS above)
and/or reversing the table order might influence results

how did you determine that mssql was using the index on table2.id and attempting to match through this?
0
 
efamilantAuthor Commented:
Using inner join syntax makes no difference in execution time.    

I provided simplified example of the query in my original post.   Here is the real query:

select doc_table.doctype,doc_table.create_date, icdtable.icdcode,code_table.token, code_table.did, code_table.pos_start,
code_table.pos_end, icdtable.icdid, icdtable.cid, icdtable.status,icdtable.kstatus from icdtable,code_table,doc_table where doc_table.acct= '2958404_017'
and doc_table.id=code_table.did and code_table.cid=icdtable.cid and icdtable.status <> 'X'

And the version of the query using inner joins:

select doc_table.doctype,doc_table.create_date, icdtable.icdcode,code_table.token, code_table.did, code_table.pos_start,
code_table.pos_end, icdtable.icdid, icdtable.cid, icdtable.status,icdtable.kstatus from doc_table INNER JOIN code_table ON doc_table.id = code_table.did
INNER JOIN icdtable ON code_table.cid =icdtable.cid  where doc_table.acct= '2958404_017' and icdtable.status <> 'X'

The execution time is identical in each case.  So whether I explicitly use inner joins in my statement or implicitly use them, as I did in the original query, the execution time is the same.
0
Configuration Guide and Best Practices

Read the guide to learn how to orchestrate Data ONTAP, create application-consistent backups and enable fast recovery from NetApp storage snapshots. Version 9.5 also contains performance and scalability enhancements to meet the needs of the largest enterprise environments.

 
PortletPaulCommented:
can you post the execution plan xml? (as a file)

regardless of execution timings - I'd advise use of ANSI syntax for maintainability if nothing else

{+edit sorry fixed spelling}
0
 
PortletPaulCommented:
perhaps try mixing it up a little too? e.g.
SELECT
      doc_table.doctype
    , doc_table.create_date
    , icdtable.icdcode
    , code_table.token
    , code_table.did
    , code_table.pos_start
    , code_table.pos_end
    , icdtable.icdid
    , icdtable.cid
    , icdtable.STATUS
    , icdtable.kstatus
FROM code_table
INNER JOIN doc_table ON code_table.did = doc_table.id
    AND doc_table.acct = '2958404_017'
INNER JOIN icdtable ON code_table.cid = icdtable.cid
    AND icdtable.STATUS <> 'X'

Open in new window

0
 
efamilantAuthor Commented:
As requested.
executionplan.txt
0
 
efamilantAuthor Commented:
If I use this version of the query, with the hint as part of it, the query executes instantly.

select doc_table.doctype,doc_table.create_date, icdtable.icdcode,code_table.token, code_table.did, code_table.pos_start,
code_table.pos_end, icdtable.icdid, icdtable.cid, icdtable.status,icdtable.kstatus from icdtable WITH (INDEX=IX_icdtable_cid),code_table,doc_table where doc_table.acct= '999999999'
and doc_table.id=code_table.did and code_table.cid=icdtable.cid and icdtable.status <> 'X'

But my question is, is there a way to do this without using hints.
0
 
PortletPaulCommented:
how current are your statistics? perhaps these should be updated?

the most costly operation in that execution plan is a clustered index seek in the doc_table searching for the '2058404_017' (node 4), and index IX_icdtable_cid is already being used (node 5).
0
 
Anthony PerkinsCommented:
As indicated previously and without looking at your Execution Plan I would be surprised if it was not as simple as some stale statistics.
0

Featured Post

Get your Disaster Recovery as a Service basics

Disaster Recovery as a Service is one go-to solution that revolutionizes DR planning. Implementing DRaaS could be an efficient process, easily accessible to non-DR experts. Learn about monitoring, testing, executing failovers and failbacks to ensure a "healthy" DR environment.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now