Performance Tuning :: Options For Optimizing SQL Running Against Wide Tables
Nov 13, 2012
The scale of the tests that generate the following scenario is not huge right now, only 50 users simulated (or you can think of them as independently running threads if you like). But here is the crunch, the queries generated (from generic transaction layer) are all running against a table that has 600 columns! We can't really control this right now, but this is causing masses amounts of IO (5GB per request) making requests queue for disk availability (which are setup RAID 0/1); its even noticable for as few as 3 threads.
I have rendered the SQL on one occasion to execute in 13 seconds for a single user but this appears short lived as when stats were freshly gathered it went up to the normal 90-120 seconds. I've added the original query to the file, however the findings here along with our DBA (who I trust implicitly) suggest that no amount of editing the query will improve the response times, increasing the PGA/SGA (currently 4/6GB respectively) will only delay the queuing for a bit and compression can work either. In short it looks as though we've hit hardware restrictions already for this particular scenario.
As I can't really explain how my rendered query no longer takes 13 seconds, it's niggling me that we might be missing a trick.So I was hoping for some guidance on possible ways of optimising these type of queries against such wide tables, in other words possibilities that we haven't considered...
I have a view, below, which does few left outer joins to the same V_MARKET view to get data i need. When I run SQL by itself, ut runs pretty fast, 2-5 seconds. But when I do "select * from V_DEPT_DATA where busines_date = '01-APR-10'", it takes more than 10 minutes to run. I added all needed indexes and still have problems with it .
CREATE OR REPLACE VIEW V_DEPT_DATA AS SELECT v1.business_date , v1.division , v1.department , v1.account , en.trader , [code]........
I am working on an assignement where client is using Oracle 10g but stuck to using RBO Now the application team, from the GUI available to them build dynamic queries and some of them run very slow.
Neither the code can not be changed to tune the queries nor do we get the exact step in the plan which is an issue (being RBO).For some long running queries the Tuning advisor is not producing any recommendations.
Another hurdle is that all the application users are using same application user id so we can not write a logon trigger to use CBO for some particular queries to see what is happening in the background!
We have a table with huge data which is skewed on a 'status' column. The 'status' column has 6 distinct values with 1 particular value occupying 80-85% records.
In the batch process we query the data on the status and process the retrieved records. My senior is insisting on partitioning which I see not much feasible considering cost implications just for a part of functionality
See there are 6 status 'A','B','C','D','E','F'
with 'A' occupying 80% records 'B' to 'F' occupies 2% till 14% records in the table(approx)
1) Create a conditional index on status (using case) to have records with all statuses except 'A' Then create If-ELSE structure
IF input parameter is 'A' select /*+ FULL Parallel(t) */ * from t where status='A'; ELSE Select /*+ INDEX (t conditional_index) */ * from t where status in ('B','C'); END IF;
I want to create conditional index here for 2 reasons
1] since it will have values for status except 'A' this nullify the chance that this index will be picked up when status='A' will be queried Thus making the performance worst (status ='A' is for 80% records) - The IF-ELSE is additional protection 2] Less impact on the DMLS as the index will not be on status='A' which contribute to large chunk of records
2)Populate a dummy table which would contain rowid and status. Since the business closes at 21:00 and batch process starts at 21:30 Between these times periods refresh the dummy table every day using merge (to catch business transactions during the day)
Now during the batch process retrieve records from the main table using the rowids in the dummy table depending on the input status value
3)Create index on status Make sure hard coded status values are used in the database procedures Gather stats with the histograms And leave it to the Optimizer to choose the best possible path
How do i find a particular SQL or a set of SQL's which are excuted against a table (user identified table) that is either a very frequently executed query against that table or high impact SQL against that table? I am currently looking through the AWR reports to go through all the queries but i was wondering if there are any dictionary views where we can find this info from?
We have a program that is taking about 13-14 hours to run and we need to generate traces to see where it is taking so long. I usually use 10046 for the tracing, I'm wondering if the traces can be built incremently so that it doesn't become one huge trace file.
I have a Query(report) which is running in <5 mins in one Scheme, where as the same is running for a long time in second schema. I have identified that an Index is scanning for more than 2000 Millions of records in second Schema, but this is scanning only 440 Millions in First Schema and hence it is fast. I am expecting the same to be done in Second schema.
I have verified the following All records in tables in 2 schemas are same. All indexes are same Analyzed the tables Gathered Histogram on all the columns as per the first schema.
But now i still have the same problem, don't know what could be the problem.
How can i check if paging happening while running the query. As i have 4gb of PGA target but the query is taking long time in parallel and has hash join.
Our application servers will be running a SELECT which returns zero rows all the time.This SELECT is put into a package and this package will be called by application servers very frequently which is causing unnecessary CPU.
Original query and plan
SQL> SELECT SEGMENT_JOB_ID, SEGMENT_SET_JOB_ID, SEGMENT_ID, TARGET_VERSION FROM AIMUSER.SEGMENT_JOBS WHERE SEGMENT_JOB_ID NOT IN (SELECT SEGMENT_JOB_ID FROM AIMUSER.SEGMENT_JOBS) 2 3 4 5 ; [code]....
Which option will be better or do we have other options?They need to pass the column's with zero rows to a ref cursor.
How can we check completion status for running sql query. i.e. how much % completed
SQL> begin 2 delete from gsmcrmdw.wc_loy_txn_f_aa 3 where integration_id in 4 ( select integration_id 5 from support_olap.recover_wc_loy_txn_f_953to955 6 ); 7 commit; 8 end; 9 /
I have to do the optimization of a query that has the following characteristics:
- Takes 3 hours to process - Performs the inner join with 30 tables - Produces an output of 280 million records with 450 fields
First of all it is not feasible to make 30 updates (one for each table) to 280 million records.
The best solution that I had found so far was to create 3 temporary tables, where each of them to do the join with 1/3 of the 30 tables, and in the end I make the join between the main table and these three tables temporary.
I know that you will ask (or maybe not) to the query and samples, but it is impossible to create 30 examples.
how to optimize this type of querys that perform the join with multiple tables and produce a large output with (too) many columns.
We have a procedure, which do truncate to some of the tables. Most of the time it finished in short of spam of time. But from last few days, it is taking much longer time.
So our situation is pretty simple. We have 3 tables.
A, B and C
the model is A->>B->>C
Currently A, B and C are range partitioned on a key created_date however it's typical that only C is every qualfied with created date. There is a foreign key from B -> A and C -> Bhave many queries where the data is identified by state that is indexed currently non partitioned on columns in A ... there are also indexes on the foreign keys that get from C -> B -> A. Again these are non partitioned indexes at this time.
It is typical that we qualifier A on either account or user or both. There are indexes (non partitioned on these) We have a problem with now because many of the queries use leading wildcards ie. account like '%ACCOUNT' etc. This often results in large full table scans. Our solution has been to remove the leading wildcard.
We are wondering how we can benefit from partitioning and or sub partitioning table A. since it's partitioned on created_date but rarely qualified by that. We are also wondering where and how we can benefit from either global partitioned index or local partitioned indexes on tables A. We suspect that the index on the foreign key from C to B could be a local partitioned index.
i am using 11.2.0.3.0 version of oracle. We are planning to move some ~40 tables/indexes to new encrypted tablespace as a part of TDE(transparent data encryption). Currently three tables are having size ~30GB and one having ~800GB other have <2GB in size. And tables/indexes are altogether placed in different tablespaces.
whether i should create as many no of encrypted table spaces as it was before as unencrypted tablespace? or I should create one encrypted tablespace and move all the tables/indexes into that?
I have tried a lot by alternate solutions like rearranging the order of tables in join and moving where conditions before but no success...Its a bottleneck and I could not have indexes on these tables in production...I want to change the approach in subquery
SELECT g.COLUMN1, g.COLUMN2, e.COLUMN3, g.COLUMN4, MIN(e.dat1) KEEP ( DENSE_RANK FIRST ORDER BY date2 Desc) * -1, min(to_char(date3,'dd-mm-yyyy')) [code]....
Ways for improving the Table performance which holds million of records for oracle. Currently we have partitioning and indexing but it doesn't seem to work.
SELECT department_id FROM (SELECT department_id FROM employees UNION SELECT department_id FROM employees_old ) WHERE department_id=100; [code]....
The index has been created on both depart_id for the two tables. The only difference between the two I observed was the 1 recursive call for the 1st sql.and also, one additional view in the plan.There is a little difference in bytes sent over the network.
I have a view on base tables holding historical data for previous 60 months(one table per month) with union all operators.create index on those base tables will improve performance or creating a primary key with disabled novalidate will improve for retrieving data?
The view has around 8 million data and used as a fact table with 4 dimension tables.A DTS package from MSSql side refreshes OLAP cube by retrieving data from these tables in oracle.
We have a huge table in production, with LONG column. We are trying to change its datatype to CLOB. The table has 120 Million records and is of 270 GB in size.
We tried using the oracle expdp/impdp option to try the conversion in our perf environment. With 32 parallels, the export completed in 1.5 hrs. However, the import took 13 hrs.
I also tried the to_lob option using inserts, it went on for 20 hrs and I killed the process. Are there any ways to improve the performance of LONG to CLOB conversion on huge tables?
Create small functional indexes for special cases in very large tables.
When there is a column having one values in 99% records and another values that have to be search for, it is possible to create an index using null value. Index will be small and the rebuild fast.
Example
create index vh_tst_decode_ind_if1 on vh_tst_decode_ind (decode(S,'I','I',null),style)
It is possible to do index more selective when the key is updated and there are many records to create more levels in b-tree.
create index vh_tst_decode_ind_if3 on vh_tst_decode_ind (decode(S,'I','I',null), decode(S,'I',style,null) )
To access the record can by like:
SQL> select --+ index(vh_tst_decode_ind_if3) 2 style ,count(*) 3 from vh_tst_decode_ind 4 where 5 decode(S,'I','I',null)='I' 6 group by style 7 ;
Looking to understand the difference between instance tuning and database tuning.
What is the difference between these two tuning exercises? I understand that an instance is memory based structures (logical) where as database consists of physical structures.
However, how does one tune a database the physical structure? Does it have to do with file placements/block sizes etc. Would you agree that a lot of that is taken care by ASM now in 11g? What tools are required/available (third party as well as oracle supplied) for these types of tuning scenarios?
I have two tables with 113M records in DWH_BILL_DET & 103M in prd_rerate_chg_que and Im running following merge query, which is running for 13 hrs to update records, which is quiet longer time.
SQL> explain plan for MERGE /*+ parallel (rq, 16) */ INTO DWH_BILL_DET rq USING (SELECT rated_que_rowid, detail_rerate_flag_code, rerate_sel_key,
How the length of column width effects index performance?
For example if i had IOT table emp_iot with columns: (id number, job varchar2(20), time date, plan number)
Table key consist of(id, job, time)
Column JOB has fixed list of distinct values ('ANALYST', 'NIGHT_WORKED', etc...).
What performance increase i could expect if in column "job" i would store not names but concrete numbers identifying job names. For e.g. i would store "1" instead 'ANALYST' and "2" instead 'NIGHT_WORKED'.