SlideShare a Scribd company logo

Using Indexed field effectively in Access Queries with Voyager

•Download as PPTX, PDF•
•
Alan Manifold
Alan Manifold

Using the indexes in the Voyager Oracle database can speed up your queries considerably. This presentation suggests ways to find out what is indexed and to use the indexes.

1 of 27
USING INDEXED FIELDS EFFECTIVELY IN ACCESS QUERIES Alan Manifold Systems Implementation Manager Purdue University Libraries European Endeavor Users Group 2005 Parallel sessions III 10 June 2005 12.00 - 12.45
Using Indexed Fields Effectively in Access Queries PHONEBOOKS Aaa—AlaA Ala—Bah A A phonebook is an index, but not the efficient kind you would want your computer to use. The basic lookup process is a scan from beginning to end, looking at the index tabs at the top of each page.
Using Indexed Fields Effectively in Access Queries AN IMPROVEMENT ABCDEFGHIJKLM NOPQRSTUVWXYZ ABCDEF GHIJKLM NOPQRS TUVWXYZ ABC DEF GHI JKLM NOP QRS TUV WXYZ JK LM J K L M N OP WX YZ W X Y Z ABCDEFGHIJKLMNOPQRSTUVWXYZ This index gets to any letter in a max of 6 tries A BC D EF G HI B C E F H I Q RS T UV O P R S U V Bah-Bur Bur-Cam
Using Indexed Fields Effectively in Access Queries EVEN BETTER In a B-Tree, each node can point to a number of child nodes, so the disk access is minimized. Nodes are (ideally) sized to make maximum use of memory, which is fast. Cam-Cop Cop-Des
Using Indexed Fields Effectively in Access Queries NOT OUR PROBLEM We don’t have to worry about how indexes work, because Oracle provides them for us. We just want to make sure we use them. Des-Die Die-Ebs
Using Indexed Fields Effectively in Access Queries WITHOUT INDEXES The alternative to using an index is a “full table scan”. This means looking at each record in the database in turn. Depending on the size of the table, this can be very slow. SELECT BIB_ID FROM BIB_TEXT WHERE TITLE LIKE "*turtles*"; “Let's see… Does record 1 have this? Nope. How about record 2? Nope. Record 3? Hmmm…” Ebs-End End-Fax
Using Indexed Fields Effectively in Access Queries IDENTIFYING INDEXES Use the query on the following slide in SQL*Plus to identify what indexes there are in the Voyager database. Fax-Fly Fly-Gba
Using Indexed Fields Effectively in Access Queries IDENTIFYING INDEXES SELECT USER_INDEXES.INDEX_NAME, USER_INDEXES.TABLE_NAME, COLUMN_POSITION, SUBSTR(COLUMN_NAME,1,40) COL_NAME FROM USER_INDEXES, USER_IND_COLUMNS WHERE USER_INDEXES.INDEX_NAME = USER_IND_COLUMNS.INDEX_NAME; Gba-God God-Hel
Using Indexed Fields Effectively in Access Queries OUR EXAMPLES TABLE_NAME INDEX_NAME # COLUMN_NAME ELINK_INDEX ELINK_INDEX_ELINK_ID 1 ELINK_ID ELINK_INDEX ELINK_INDEX_LINK_NORMAL 1 LINK_TEXT_NORMAL ELINK_INDEX ELINK_INDEX_RECID_RECTYPE_IDX 1 RECORD_ID ELINK_INDEX ELINK_INDEX_RECID_RECTYPE_IDX 2 RECORD_TYPE FINE_FEE FINE_FEE_IDX 1 FINE_FEE_ID FINE_FEE FINE_FEE_IDX1 1 ITEM_ID FINE_FEE FINE_FEE_IDX1 2 FINE_FEE_TYPE FINE_FEE FINE_FEE_IDX1 3 PATRON_ID FINE_FEE FINE_FEE_IDX1 4 FINE_FEE_BALANCE FINE_FEE FINE_FEE_IDX1 5 FINE_FEE_ID FINE_FEE FINE_FEE_PTN_IDX 1 PATRON_ID FUND FUND_IDX 1 LEDGER_ID FUND FUND_IDX 2 FUND_ID FUND FUND_X_CODE 1 NORMAL_FUND_CODE FUND FUND_X_NAME 1 NORMAL_FUND_NAME FUND FUND_X_PARENT 1 LEDGER_ID FUND FUND_X_PARENT 2 PARENT_FUND We’ll use these records to learn how to read the results. Hel-Hos Hos-Ico
Using Indexed Fields Effectively in Access Queries THE SIMPLEST CASE In this simplest case, the index covers only one field. The COLUMN_POSITION field is 1, and there is no 2. If you add criteria to the ELINK_ID field or link from this table to another by it, the index will be used and access will be fast. TABLE_NAME INDEX_NAME # COLUMN_NAME ELINK_INDEX ELINK_INDEX_ELINK_ID 1 ELINK_ID Ico-Inc Inc-Jar
Using Indexed Fields Effectively in Access Queries A BIT HARDER TABLE_NAME INDEX_NAME # COLUMN_NAME FUND FUND_IDX 1 LEDGER_ID FUND FUND_IDX 2 FUND_ID In this case, you can see there is a COLUMN_POSITION (#) field value of 2, as well as one of 1. If you search this field by FUND_ID without LEDGER_ID, it will do a full table scan. You must have the field with position 1 before it can use the index for the field in position 2. Jar-Jun Jun-Kaa
Using Indexed Fields Effectively in Access Queries THESE CAN’t USE IT … Because neither of these two queries refers to the LEDGER_ID, they can’t use the FUND_IDX index. They’ll have to do a full scan of the FUND table. Kaa-Kri Kri-Lap
Using Indexed Fields Effectively in Access Queries … BUT THESE CAN These two queries could both use the FUND_IDX index, as they use the LEDGER_ID as well as the FUND_ID Lap-Lib Lib-Llo
Using Indexed Fields Effectively in Access Queries A HARDER ONE YET TABLE_NAME INDEX_NAME # COLUMN_NAME FINE_FEE FINE_FEE_IDX1 1 ITEM_ID FINE_FEE FINE_FEE_IDX1 2 FINE_FEE_TYPE FINE_FEE FINE_FEE_IDX1 3 PATRON_ID FINE_FEE FINE_FEE_IDX1 4 FINE_FEE_BALANCE FINE_FEE FINE_FEE_IDX1 5 FINE_FEE_ID To use this index, you need an ITEM_ID, then a FINE_FEE_TYPE, then a PATRON_ID, then the FINE_FEE_BALANCE, then the FINE_FEE_ID. You don’t have to use all of these fields, but you must use at least the ones with smaller numbers than the field you’re interested in. Llo-Max Max-Moo
Using Indexed Fields Effectively in Access Queries USING THE INDEX This query can use the FINE_FEE_IDX1 index since the first three fields in the index (ITEM_ID, FINE_FEE_TYPE, and PATRON_ID) are all being used for links. Moo-Nal Nal-Not
Using Indexed Fields Effectively in Access Queries CRITERIA NOTES Exact criteria such as: “reference” and “ref main” Or “ref branch” appear to work faster than fuzzy criteria such as: Like “ref*” Indexes are left- anchored, so criteria such as: Like “*refe” can’t use the indexes Not-Oax Oax-Ole
Using Indexed Fields Effectively in Access Queries THE FUDGE FACTOR Where I have shown examples of queries, I have said the query “could” or “can” use the index. Whether it does or not is up to Oracle and the ODBC drivers. Ole-Opa Opa-Pac
Using Indexed Fields Effectively in Access Queries EXAMPLE #1 Let’s look at the indexes related to this query and see how good it is. Pac-Pez Pez-Pur
Using Indexed Fields Effectively in Access Queries THE LINKS WE USED BIB_TEXT BIB_TEXT_BIB_ID_IDX 1 BIB_ID LOCATION LOCATION_LOC_ID_IDX 1 LOCATION_ID BIB_MFHD BIB_MFHD_BIBID_MFHDID_IDX 1 BIB_ID BIB_MFHD BIB_MFHD_BIBID_MFHDID_IDX 2 MFHD_ID BIB_MFHD BIB_MFHD_MFHDID_BIBID_IDX 1 MFHD_ID BIB_MFHD BIB_MFHD_MFHDID_BIBID_IDX 2 BIB_ID Most of the links we used were on indexed fields. MFHD_MASTER MFHD_MASTER_MFHD_ID_IDX 1 MFHD_ID MFHD_ITEM MFHD_ITEM_ITEM_IDX 1 ITEM_ID MFHD_ITEM MFHD_ITEM_MFHD_IDX 1 MFHD_ID ITEM ITEM_IDX 1 ITEM_ID ITEM ITEM_TEMPLOC_PERMLOC_IDX 1 TEMP_LOCATION ITEM ITEM_TEMPLOC_PERMLOC_IDX 2 PERM_LOCATION Pur-Qua Qua-Quo
Using Indexed Fields Effectively in Access Queries THE CRITERIA WE USED MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 1 NORMALIZED_CALL_NO MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 2 CALL_NO_TYPE MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 3 LOCATION_ID MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 4 SUPPRESS_IN_OPAC MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 5 MFHD_ID MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 6 DISPLAY_CALL_NO MFHD_MASTER MFHD_MASTER_TYPE_IDX 1 CALL_NO_TYPE MFHD_MASTER MFHD_MASTER_TYPE_IDX 2 MFHD_ID CALL_NO_TYPE is indexed, but NORMALIZED_CALL_NO would be a better choice for criteria than DISPLAY_CALL_NO. In the LOCATION table, only the LOCATION_ID is indexed. Quo-Ram Ram-Rev
Using Indexed Fields Effectively in Access Queries DOES LOCATION MATTER? Although neither the LOCATION_CODE nor the LOCATION_NAME are indexed in LOCATION, it makes very little difference in speed. The shorter field (LOCATION_CODE) is preferable for criteria. In a B-Tree structure, the nodes are relatively large, so they have lots of records in them. For a small table, a full- table scan is equivalent to an index lookup. Rev-Rip Rip-Ror
Using Indexed Fields Effectively in Access Queries EXAMPLE #2 Ror-Rug Rug-San
Using Indexed Fields Effectively in Access Queries ANALYSIS All the links are on indexed fields except to PO_STATUS and INVOICE_STATUS, which are too small to count. The criterion on the INVOICE_STATUS_DESC field from the INVOICE_STATUS table is okay, too. San-Sol Sol-Sun
Using Indexed Fields Effectively in Access Queries EXAMPLE #3 BIB_INDEX BIB_INDEX_BIB_ID_IDX 1 BIB_ID BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 1 INDEX_CODE BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 2 NORMAL_HEADING BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 3 BIB_ID BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 4 DISPLAY_HEADING To use BIB_INDEX to get all titles by an author (100 or 700 fields), specify the INDEX_CODE field as well as the NORMAL_HEADING field. 4 secs >5 mins 0 secs Sun-Tee Tee-Tod
Using Indexed Fields Effectively in Access Queries EXAMPLE #4 There’s no help for this query. It needs to use criteria that aren’t indexed anywhere. It will take a long time to run no matter what. Tod-Voy Voy-Wah
Using Indexed Fields Effectively in Access Queries EXAMPLE #5 Not only does this query not use the normalized last name for its criterion, it also requires the user to enter the names in mixed case. Wah-Who Who-You
Using Indexed Fields Effectively in Access Queries SUMMARY Using indexed fields in queries is mostly a matter of common sense, and is not difficult. But it can make a major difference in the speed of queries. You-Zen Zen-Zzy

More Related Content

Using Indexed field effectively in Access Queries with Voyager

  • 1. USING INDEXED FIELDS EFFECTIVELY IN ACCESS QUERIES Alan Manifold Systems Implementation Manager Purdue University Libraries European Endeavor Users Group 2005 Parallel sessions III 10 June 2005 12.00 - 12.45
  • 2. Using Indexed Fields Effectively in Access Queries PHONEBOOKS Aaa—AlaA Ala—Bah A A phonebook is an index, but not the efficient kind you would want your computer to use. The basic lookup process is a scan from beginning to end, looking at the index tabs at the top of each page.
  • 3. Using Indexed Fields Effectively in Access Queries AN IMPROVEMENT ABCDEFGHIJKLM NOPQRSTUVWXYZ ABCDEF GHIJKLM NOPQRS TUVWXYZ ABC DEF GHI JKLM NOP QRS TUV WXYZ JK LM J K L M N OP WX YZ W X Y Z ABCDEFGHIJKLMNOPQRSTUVWXYZ This index gets to any letter in a max of 6 tries A BC D EF G HI B C E F H I Q RS T UV O P R S U V Bah-Bur Bur-Cam
  • 4. Using Indexed Fields Effectively in Access Queries EVEN BETTER In a B-Tree, each node can point to a number of child nodes, so the disk access is minimized. Nodes are (ideally) sized to make maximum use of memory, which is fast. Cam-Cop Cop-Des
  • 5. Using Indexed Fields Effectively in Access Queries NOT OUR PROBLEM We don’t have to worry about how indexes work, because Oracle provides them for us. We just want to make sure we use them. Des-Die Die-Ebs
  • 6. Using Indexed Fields Effectively in Access Queries WITHOUT INDEXES The alternative to using an index is a “full table scan”. This means looking at each record in the database in turn. Depending on the size of the table, this can be very slow. SELECT BIB_ID FROM BIB_TEXT WHERE TITLE LIKE "*turtles*"; “Let's see… Does record 1 have this? Nope. How about record 2? Nope. Record 3? Hmmm…” Ebs-End End-Fax
  • 7. Using Indexed Fields Effectively in Access Queries IDENTIFYING INDEXES Use the query on the following slide in SQL*Plus to identify what indexes there are in the Voyager database. Fax-Fly Fly-Gba
  • 8. Using Indexed Fields Effectively in Access Queries IDENTIFYING INDEXES SELECT USER_INDEXES.INDEX_NAME, USER_INDEXES.TABLE_NAME, COLUMN_POSITION, SUBSTR(COLUMN_NAME,1,40) COL_NAME FROM USER_INDEXES, USER_IND_COLUMNS WHERE USER_INDEXES.INDEX_NAME = USER_IND_COLUMNS.INDEX_NAME; Gba-God God-Hel
  • 9. Using Indexed Fields Effectively in Access Queries OUR EXAMPLES TABLE_NAME INDEX_NAME # COLUMN_NAME ELINK_INDEX ELINK_INDEX_ELINK_ID 1 ELINK_ID ELINK_INDEX ELINK_INDEX_LINK_NORMAL 1 LINK_TEXT_NORMAL ELINK_INDEX ELINK_INDEX_RECID_RECTYPE_IDX 1 RECORD_ID ELINK_INDEX ELINK_INDEX_RECID_RECTYPE_IDX 2 RECORD_TYPE FINE_FEE FINE_FEE_IDX 1 FINE_FEE_ID FINE_FEE FINE_FEE_IDX1 1 ITEM_ID FINE_FEE FINE_FEE_IDX1 2 FINE_FEE_TYPE FINE_FEE FINE_FEE_IDX1 3 PATRON_ID FINE_FEE FINE_FEE_IDX1 4 FINE_FEE_BALANCE FINE_FEE FINE_FEE_IDX1 5 FINE_FEE_ID FINE_FEE FINE_FEE_PTN_IDX 1 PATRON_ID FUND FUND_IDX 1 LEDGER_ID FUND FUND_IDX 2 FUND_ID FUND FUND_X_CODE 1 NORMAL_FUND_CODE FUND FUND_X_NAME 1 NORMAL_FUND_NAME FUND FUND_X_PARENT 1 LEDGER_ID FUND FUND_X_PARENT 2 PARENT_FUND We’ll use these records to learn how to read the results. Hel-Hos Hos-Ico
  • 10. Using Indexed Fields Effectively in Access Queries THE SIMPLEST CASE In this simplest case, the index covers only one field. The COLUMN_POSITION field is 1, and there is no 2. If you add criteria to the ELINK_ID field or link from this table to another by it, the index will be used and access will be fast. TABLE_NAME INDEX_NAME # COLUMN_NAME ELINK_INDEX ELINK_INDEX_ELINK_ID 1 ELINK_ID Ico-Inc Inc-Jar
  • 11. Using Indexed Fields Effectively in Access Queries A BIT HARDER TABLE_NAME INDEX_NAME # COLUMN_NAME FUND FUND_IDX 1 LEDGER_ID FUND FUND_IDX 2 FUND_ID In this case, you can see there is a COLUMN_POSITION (#) field value of 2, as well as one of 1. If you search this field by FUND_ID without LEDGER_ID, it will do a full table scan. You must have the field with position 1 before it can use the index for the field in position 2. Jar-Jun Jun-Kaa
  • 12. Using Indexed Fields Effectively in Access Queries THESE CAN’t USE IT … Because neither of these two queries refers to the LEDGER_ID, they can’t use the FUND_IDX index. They’ll have to do a full scan of the FUND table. Kaa-Kri Kri-Lap
  • 13. Using Indexed Fields Effectively in Access Queries … BUT THESE CAN These two queries could both use the FUND_IDX index, as they use the LEDGER_ID as well as the FUND_ID Lap-Lib Lib-Llo
  • 14. Using Indexed Fields Effectively in Access Queries A HARDER ONE YET TABLE_NAME INDEX_NAME # COLUMN_NAME FINE_FEE FINE_FEE_IDX1 1 ITEM_ID FINE_FEE FINE_FEE_IDX1 2 FINE_FEE_TYPE FINE_FEE FINE_FEE_IDX1 3 PATRON_ID FINE_FEE FINE_FEE_IDX1 4 FINE_FEE_BALANCE FINE_FEE FINE_FEE_IDX1 5 FINE_FEE_ID To use this index, you need an ITEM_ID, then a FINE_FEE_TYPE, then a PATRON_ID, then the FINE_FEE_BALANCE, then the FINE_FEE_ID. You don’t have to use all of these fields, but you must use at least the ones with smaller numbers than the field you’re interested in. Llo-Max Max-Moo
  • 15. Using Indexed Fields Effectively in Access Queries USING THE INDEX This query can use the FINE_FEE_IDX1 index since the first three fields in the index (ITEM_ID, FINE_FEE_TYPE, and PATRON_ID) are all being used for links. Moo-Nal Nal-Not
  • 16. Using Indexed Fields Effectively in Access Queries CRITERIA NOTES Exact criteria such as: “reference” and “ref main” Or “ref branch” appear to work faster than fuzzy criteria such as: Like “ref*” Indexes are left- anchored, so criteria such as: Like “*refe” can’t use the indexes Not-Oax Oax-Ole
  • 17. Using Indexed Fields Effectively in Access Queries THE FUDGE FACTOR Where I have shown examples of queries, I have said the query “could” or “can” use the index. Whether it does or not is up to Oracle and the ODBC drivers. Ole-Opa Opa-Pac
  • 18. Using Indexed Fields Effectively in Access Queries EXAMPLE #1 Let’s look at the indexes related to this query and see how good it is. Pac-Pez Pez-Pur
  • 19. Using Indexed Fields Effectively in Access Queries THE LINKS WE USED BIB_TEXT BIB_TEXT_BIB_ID_IDX 1 BIB_ID LOCATION LOCATION_LOC_ID_IDX 1 LOCATION_ID BIB_MFHD BIB_MFHD_BIBID_MFHDID_IDX 1 BIB_ID BIB_MFHD BIB_MFHD_BIBID_MFHDID_IDX 2 MFHD_ID BIB_MFHD BIB_MFHD_MFHDID_BIBID_IDX 1 MFHD_ID BIB_MFHD BIB_MFHD_MFHDID_BIBID_IDX 2 BIB_ID Most of the links we used were on indexed fields. MFHD_MASTER MFHD_MASTER_MFHD_ID_IDX 1 MFHD_ID MFHD_ITEM MFHD_ITEM_ITEM_IDX 1 ITEM_ID MFHD_ITEM MFHD_ITEM_MFHD_IDX 1 MFHD_ID ITEM ITEM_IDX 1 ITEM_ID ITEM ITEM_TEMPLOC_PERMLOC_IDX 1 TEMP_LOCATION ITEM ITEM_TEMPLOC_PERMLOC_IDX 2 PERM_LOCATION Pur-Qua Qua-Quo
  • 20. Using Indexed Fields Effectively in Access Queries THE CRITERIA WE USED MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 1 NORMALIZED_CALL_NO MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 2 CALL_NO_TYPE MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 3 LOCATION_ID MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 4 SUPPRESS_IN_OPAC MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 5 MFHD_ID MFHD_MASTER MFHD_MASTER_NORM_TYPE_LOC_IDX 6 DISPLAY_CALL_NO MFHD_MASTER MFHD_MASTER_TYPE_IDX 1 CALL_NO_TYPE MFHD_MASTER MFHD_MASTER_TYPE_IDX 2 MFHD_ID CALL_NO_TYPE is indexed, but NORMALIZED_CALL_NO would be a better choice for criteria than DISPLAY_CALL_NO. In the LOCATION table, only the LOCATION_ID is indexed. Quo-Ram Ram-Rev
  • 21. Using Indexed Fields Effectively in Access Queries DOES LOCATION MATTER? Although neither the LOCATION_CODE nor the LOCATION_NAME are indexed in LOCATION, it makes very little difference in speed. The shorter field (LOCATION_CODE) is preferable for criteria. In a B-Tree structure, the nodes are relatively large, so they have lots of records in them. For a small table, a full- table scan is equivalent to an index lookup. Rev-Rip Rip-Ror
  • 22. Using Indexed Fields Effectively in Access Queries EXAMPLE #2 Ror-Rug Rug-San
  • 23. Using Indexed Fields Effectively in Access Queries ANALYSIS All the links are on indexed fields except to PO_STATUS and INVOICE_STATUS, which are too small to count. The criterion on the INVOICE_STATUS_DESC field from the INVOICE_STATUS table is okay, too. San-Sol Sol-Sun
  • 24. Using Indexed Fields Effectively in Access Queries EXAMPLE #3 BIB_INDEX BIB_INDEX_BIB_ID_IDX 1 BIB_ID BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 1 INDEX_CODE BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 2 NORMAL_HEADING BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 3 BIB_ID BIB_INDEX BIB_INDEX_CODE_NORM_DISP_IDX 4 DISPLAY_HEADING To use BIB_INDEX to get all titles by an author (100 or 700 fields), specify the INDEX_CODE field as well as the NORMAL_HEADING field. 4 secs >5 mins 0 secs Sun-Tee Tee-Tod
  • 25. Using Indexed Fields Effectively in Access Queries EXAMPLE #4 There’s no help for this query. It needs to use criteria that aren’t indexed anywhere. It will take a long time to run no matter what. Tod-Voy Voy-Wah
  • 26. Using Indexed Fields Effectively in Access Queries EXAMPLE #5 Not only does this query not use the normalized last name for its criterion, it also requires the user to enter the names in mixed case. Wah-Who Who-You
  • 27. Using Indexed Fields Effectively in Access Queries SUMMARY Using indexed fields in queries is mostly a matter of common sense, and is not difficult. But it can make a major difference in the speed of queries. You-Zen Zen-Zzy