calculate statistical performance problems

Hi all

We run statistical calculation on a table that is completely deleted daily, filled and then analyzed. However, the statistical component of calculation takes more and more time each day. The table retains a kind of diary/history of statistics that were calculated before the deleted data? Or is there a reason why statistical calculation on about the same amount of data every day should take longer and longer time? We took 10g.

A few notes:
I can't truncate table because I need to restore in case of problems. I could drop and recreate once if it helps.
Before deleting and filling, I disable all indexes, and then re-create them.
Transfer us about 10 million rows every day in this table.
A year of the entire process (fill + analyze) has taken 3 hours, now it takes 18 hours.
We also got this error on the table: ORA-01555: snapshot too old: rollback segment number with the name ° ORA-02063 too small: before the T_SRC line

This is the script in the procedure:

    execute immediate 'alter index IND_TAB1 unusable';

    delete from TAB1;

    insert /*+ APPEND */ into TAB1
    select
      COL1
     ,COL2
      from V_SRC1
     where change_date   between sysdate-1 and sysdate
        or creation_date between sysdate-1 and sysdate;

    execute immediate 'alter index IND_TAB1  rebuild';

    execute immediate 'analyze table TAB1 compute statistics';

V_SRC1 estimated T_SRC table on the remote system.

Thank you for the ideas.

Hello

When you insert and remove a large amount of data in a table, it breaks and he could accumulate a significant amount of unused disk space. When you perform a full table scan, Oracle scans the segment of the upper limit (HWM) that points to the location of the block used above, i.e. it scans all empty blocks as well. In order to reset the HWM, you can issue ALTER TABLE MOVE. Despite what the name suggest command, your table will not be moved anywhere, it will simply rebuild and recover any unused free space. If there are all the indexes defined on the table, they would have to rebuild after that as well (ALTER INDEX REBUILD of ).

Alternatively, you may consider partitioning by date range (if your license allows it). With partitioning everything would become much simpler:

(1) you load your data into the table and everything check
(2) if it is ok, you just drop the previous partition; Otherwise, restore.

With regard to the CALCULATION - it goes through all the data to calculate statistics, instead of using a small sample (in general, a few percentage points). It is slower than the ESTIMATE, and in most cases does not provide any advantage over it.

DBMS_STATS ANALYZE vs - using ANALYZE for the collection of statistics concerning is now obsolete and not recommended. DBMS_STATS offers the same features, and much more, given that your version is 10g you should use DBMS_STATS and not to ANALYZE.

Best regards
Nikolai

Tags: Database

Similar Questions

EBS 11i performance problem
Hi hussein, helios, & all,

Happy New Year to you :)
I hope we have a better year 2011 year of the rabbit.
I hope that we have even better in 2011.

I just want to learn something from each of you through the year 2010.

Can you share how tackle you performance problems in EBS?
I know that there is joint action, that you have done to improve performance.
It should be after all statistical analyses of performance as Statspack, SAR,
and other output of the performance monitoring report.

What actions / solutions did you resolve performance issues in EBS?

Is it upgraded the server itself for a more powerful?

For example, our users are complaining about the Oracle Apps very slow race at any time of the day. This produces 20% of the day(performance problem) and 80%, as the Oracle applications is running fine (no complaints of users).
The appearance of performance problem is not fixed in time, sometimes in the morning and in the afternoon sometimes. Sometimes, even when there are a few users logged in and sometimes it is no problem of performance, even if many users are connected. If you run a statspack you see programs to eat a lot of resources, but they are oracle programs if you can not touch or tune right? So to keep users calm, they agreed to get rebooted server. Temporarily, this solves the performance problem. So we just keep it this way. My suspect is that there are many programs of concurrent background (on request) who is running who is right for which users get slow even if only a few is connected to it. I want to know what are simultaneous programs running at the time where the performance issue of gets users.
Is there a sqlplus program to list all concurrent programs running?

Thank you very much

Mrs. Mina
What actions / solutions did you resolve performance issues in EBS?

Depends on what kind of performance issues you have.

Is it upgraded the server itself for a more powerful?

Not necessarily.

For example, our users are complaining about the Oracle Apps very slow race at any time of the day. This produces 20% of the day(performance problem) and 80%, as the Oracle applications is running fine (no complaints of users).

What part of the application? What activities do you have when there are performance issues?

The appearance of performance problem is not fixed in time, sometimes in the morning and in the afternoon sometimes. Sometimes, even when there are a few users logged in and sometimes it is no problem of performance, even if many users are connected. If you run a statspack you see programs to eat a lot of resources, but they are oracle programs if you can not touch or tune right? So to keep users calm, they agreed to get rebooted server. Temporarily, this solves the performance problem. So we just keep it this way.

Please see old similar threads.

Performance optimization
http://forums.Oracle.com/forums/search.jspa?threadID=&q=performance+tuning&objid=C3&DateRange=all&userid=&NumResults=15&rankBy=10001

My suspect is that there are many programs of concurrent background (on request) who is running who is right for which users get slow even if only a few is connected to it. I want to know what are simultaneous programs running at the time where the performance issue of gets users.
Is there a program of sqlplus to the list of current jobs running?

Please see these documents.

How to find database Session & process associated with a competitor, which is currently running program. [735119.1 ID]
How to extract information from SID to apply for operation [ID 280391.1]
How to find abandoned UNIX PID for a competing program [154368.1 ID]

Thank you
Hussein

a defective battery in a tablet will cause performance problems of speed

a defective battery in a tablet will cause a matter of output speed

Hello

Sometimes it can cause a performance problem, visit the following link to monitor how to take care of the battery:

Taking care of your laptop battery

http://Windows.Microsoft.com/en-us/Windows7/taking-care-of-your-laptop-battery

In order to increase the performance of Windows XP please visit the link below:

Windows XP performance

http://TechNet.Microsoft.com/en-us/library/bb457057.aspx

Reference: PC slow? Optimize your computer for peak performance

http://www.Microsoft.com/athome/Setup/optimize.aspx

Note: as long, the Tablet is connected to a power source it shouldn't affect the performance in itself. However, bad battery itself a performance problem.

WRT54GS - wireless performance problem

Dear someone

I am currently having performance problems with my WRT54GSv1.1

Just received my new internet connection to 30Mbit. Unfortunately, I am not able to download at that speed. If I connect via one of the linksys router's ethernet ports, I'm a happy person, but using the wireless makes me a little less happy.

Wireless: 15Mbit/s

Through the cable: 27Mbit/s

Network information:

The distance between the router and the computer laptop 3-4 meters

Linksys WRT54GSv1.1 - Firmware Version: v4.71.4

Network wireless - g only mode

TV - 6 (because all the surrounding neighbor use 10-11, I noticed using Cain Abel & wireless discovered 4.9.29)

Security: WPA2-Personal (TKIP + AES)

All other settings are default (just reset to the factory settings).

Laptop:

Windows Vista Business SP1 32-bit

Intel PRO/Wireless 3945ABG (installed the latest drivers from the website of intel, default settings)

When you check the diagnostic of Intel tools, it tells me the link is 54 megabits and pushes meter packages of 54 megabits, while all lower speeds remain the same.

More diagnostic information:

Percent missed beacons: 0

Percent transmit errors: 18

Current Tx power: ~ 32 mW (100%)

Supported power levels: 1.0 mW - 32.0 mW.

Hope someone can help me here.

Thanks in advance!

Kind regards

Ski Klesman

Although wireless g "connects" to 54 Mbit/s, the maximum possible wireless "data rate" (in ideal laboratory conditions) is only about 20 to 25 Mbit/s. Unfortunately, the phrase "in ideal laboratory conditions" generally excludes home intrusion!

The aerial transmission for wireless connections is higher than for wired connections. Thus, most individuals find that their Wi - Fi connection works at 50% to 70% of their speed of wired (LAN). Your wired LAN being 27 Mbps connection speed, your 15 Mbps wireless speed is within normal limits.

You might be able to tweak a bit more speed from your wireless network by optimizing all your wireless settings. Here are a few suggestions:

I guess you want actually to WPA2 encryption. If so, set to AES only. When you set the router to TKIP and AES, you actually tell the router to accept a WPA or WPA2 connection.

Also, give your network a unique SSID. Do not use "linksys". If you use "linksys", you can try to connect the router to your neighbor. Also set 'SSID Broadcast' to 'active '. This will help your computer to find and lock on the signal from your router.

Bad wireless connections are often caused by interference from other 2.4 GHz devices. This includes cordless phones, baby monitor wireless, microwave ovens, wireless mice and keyboards, wireless speakers and wireless network from your neighbor. In rare cases, Bluetooth devices can interfere. Even some 5 + GHz phones also use the 2.4 Ghz band. disconnect these devices and see if that solves your problem.

In your router, try another channel. There are 11 channels in the band of 2.4 GHz channel 1, 6 or 11 generally works better. Discover your neighbors and see what channel they use. Because the channels overlap, try to stay at least + 5 or - 5 channels of your more powerful neighbors. For example, if you have a powerful neighbour on channel 9, try any channel 1 to 4.

Also, try putting the router about 4 to 6 feet above the ground in an open area. Do not place behind your screen or other computer equipment or speakers. The antenna must be vertical.

In addition, in the computer, go to your wireless software and go to 'Favorite networks' (sometimes called 'Profiles'). There are probably a few listed networks. Remove any network called "linksys". Also remove any network that you don't recognize or that you no longer use. If your current network is not listed, enter its information (SSID, encryption (if any) and key (if any)). Select your current network and make your network by default, then set it to auto login. You may need to go to 'settings' to do this, or you may need to right click on your network and select 'Properties' or 'settings '.

If you continue to have problems, try the following:

For wireless g routers, try setting the "baud rate" at 54 Mbps.

If you still have problems, download and install the latest firmware for your router. After an update of the firmware, you must reset the default router, and then configure the router again from scratch. If you have saved a router configuration file, DO NOT use it.

I hope this helps.

4.2 APEX performance problem

Hi all

We are facing the problem of performance when we are by selecting one of the underside of responsibility that rests in the Organization as if we select APEX_01 Reporting ALL then in this case, we can see all the organization such as ETL, BRN, CHICKEN ect in home page, but I am facing performance problem when it forms that select area of responsibility to the home page. Please indicate how to check this page's performance in the APEX and also advise how to solve this problem.

LnTInfotech wrote:

so you want to say below takes more time and we need to address this request?

15.34901 224.01663 ... Run the statement: SELECT DISTINCT papf.full_name a, papf.full_name b
OF po_agents pa
per_all_people_f women's wear
org_organization_definitions org
WHERE 1 = 1
AND papf.person_id = pa.agent_id
AND org.organization_CODE = NVL(:P1_WARE_HOUSE,org.organization_CODE)
AND TRUNC (SYSDATE) BETWEEN papf.effective_start_date AND papf.effective_end_date
AND EXISTS SELECT (SEPARATE 1
OF po_headers_all poh
WHERE poh.agent_id = pa.agent_id
AND poh.org_id = org.operating_unit)

This request seems to be missing a join. An obligation to use DISTINCT in application code usually indicates that there is something seriously wrong with the data model or query...

Oracle 9i Java component performance problems

Hello
I am reorganizing a java component inherited because of performance problems.

A java stored procedure is used to trigger a shell script that runs a java component. The component connects to a series of remote directories via ftp, gets the files and load them using slq * charger.
Is there a better way to do this? I saw a few articles talking using a FTP interface directly from PL - SQL jump the java component entirely. It would be preferable to the current solution its?
Thanks in advance,
Pedro

I am reorganizing a java component inherited because of performance problems.

The first step is to identify what are the problems of performance, where they occur, and what causes them.

View details on

1. WHAT you do

2. HOW to

3. WHAT results you get

4 what ARE the results you expect to get.

ViewObject range Paging performance problem

Hi all
I am facing a performance problem with the implementation of an obligation to programmatically add a number of extra where the parameters of the clause (using bind) variable in combination with range paging.
My code looks like this
```
...

ApplicationModule am = Configuration.createRootApplicationModule("services.DossierAM", "DossierAMLocal");
ViewObject vo = am.findViewObject("DossierListView");

// apply programmatic view criteria
ViewCriteria vc = vo.createViewCriteria();
ViewCriteriaRow vcr = vc.createViewCriteriaRow();
vcr.setAttribute("Reference", "15/%");
vc.addElement(vcr);
vo.applyViewCriteria(vc, true);


// enable range paging
vo.setAccessMode(RowSet.RANGE_PAGING);
vo.setIterMode(RowIterator.ITER_MODE_LAST_PAGE_PARTIAL);
vo.setRangeSize(50);
vo.scrollToRangePage(5); // Cause a java.sql.SQLException: Parameter IN or OUT missing for index.....debugging learned that the :vc_temp_1 bind variable is not filled
// vo.scrollToRange(250); // Cause a java.sql.SQLException: Parameter IN or OUT missing for index.....debugging learned that the :vc_temp_1 bind variable is not filled

... 
  ...
```
I found 2 solutions, but they both require an application of additional database that is, performance wise, is not acceptable.
The first solution is to slip into an additional call to exectueQuery() before the call to function scrollToRangePage (int) or scrollToRange (int).
The second solution is to use the method (int) setRangeStart instead of variants scrollToRange (Page). This method performs also 2 database calls.
My question to you:
Is there another way to satisfy the requirement of programming add a certain number of parameters of the additional where clause (using the variable binding) in combination with the pagination of the range without the need to perform queries of database 2?
The code is tested with JDeveloper, 11.1.2.4.0, and 12.1.3.0.0 and behaves the same on both versions.
Kind regards
Steven.

Have you tried to create truly VC with bind variable (rather than use binding implied var created by frame)?

Something like: http://www.jobinesh.com/2010/10/creating-view-criteria-having-bind.html

Dario

Performance problems since the update of 2015.6

Hello together,
Since I updated lightroom yesterday to version 2015.6 the performance on my machine is incredible bad.
My system:
-MacBook Pro (retina, mid-2012)
-2.3 GHz Intel Core i7
-16 GB 1600 MHz DDR3
-NVIDIA GeForce GT 650M 1024Mo
-OS X El Capitan (10.11.5)
Lightroom synchronizes actually 1400 photos, but even if I stop this process, LR uses CPU so I can't work in other programs like Photoshop (updated yesterday, too). I detected this performance problem as PS freezes when I used the Healing Brush tool.
I don't know what's happening LR, but is not ideal for my workflow.
Does anyone have an idea?
If I close LR I can work in PS, even though the Healing Brush tool is slower than in the version before. But I can't always close and open LR.
Thank you
regards Denis

There are already dozens of threads discussing the same question in this forum as well as in the official comments forum here: https://feedback.photoshop.com/photoshop_family/categories/photoshop_family_photoshop_ligh troom

You can restore 6.5.1 if you wish. Here's how: How do I roll back to Lightroom 2015.1.1 or Lightroom 6.1.1? Lightroom Queen

Just replace 6.1.1 with 6.5.1

HP Pavilion 15-n005sg: performance problems and worsens

Hello

I bought this PC here is a year, has been a very good PC with good performance, but has been doing more and more slow to the month, for example, I used to run Diablo 3 on average - high specification with 60 fps constant with no performance problems, but that in the end, I can't still running at ultra low care without screen freezes and fps drops to 1-2 as well as the League of legends and other games.

It takes more and more time for my pc to turn on, which was like 30 ~ 40 seconds not takes like 3 minutes.

the performance gets really bad and should not happen on a pc of 1 year.

What can I do?

An update of the system may be your best course of action.

Your data, files, etc. will be retained. Any Windows apps store is kept.

Any program that came pre-installed is preserved.

BUT all programs which are not in the above categories will have to be resettled; your games, printer, etc.

Windows 8 system Refresh

If you found my answer helpful please say thank you by clicking the s cursor Up icon. Thank you!

Performance problem - oracle 10.2.0.4

HI Experts,
Today I met a performance problem. I pulled a simple query to get the number of records in a table existed in the databases of PROD and QA. The two databases are the same settings at parameter level. But when I launch the application, Prod database takes 10 minutes to get the result. Not sure why prod base it takes a lot of time.
Additional information:
---------------------
The number of lines is almost the same on the two databases
OPERATING SYSTEM: IBM AIX
Oracle: 10.2.0.4
QA Database - explain plan
==========================
06:40:49 SQL > select count (*) in the Siebel.s_prod_baseline;
COUNT (*)
----------
42146408
Elapsed time: 00:00:45.35 == > just 45 seconds
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Hash value of plan: 645963650
-------------------------------------------------------------------------------
| ID | Operation | Name               | Lines | Cost (% CPU). Time |
-------------------------------------------------------------------------------
|   0 | SELECT STATEMENT |                    |     1. 1165 (1) | 00:00:14 |
|   1. GLOBAL TRI |                    |     1.            |          |
|   2.   INDEX SCAN FULL | S_PROD_BASELINE_P1 |    42 M | 1165 (1) | 00:00:14 |
-------------------------------------------------------------------------------
Database of PROD explain plan
======================
06:42:59 SQL > select count (*) in the Siebel.s_prod_baseline;
COUNT (*)
----------
42730261
Elapsed time: 00:10:13.43 == > took more than 10 minutes
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Hash value of plan: 645963650
-------------------------------------------------------------------------------
| ID | Operation | Name               | Lines | Cost (% CPU). Time |
-------------------------------------------------------------------------------
|   0 | SELECT STATEMENT |                    |     1. 4160 (1) | 00:00:50 |
|   1. GLOBAL TRI |                    |     1.            |          |
|   2.   INDEX SCAN FULL | S_PROD_BASELINE_P1 |    42 M | 4160 (1) | 00:00:50 |
-------------------------------------------------------------------------------
Could if it you please let me know why oracle it takes a lot of time prod database?
Thank you..

> There was another option to check why prod database behaves differently?

Yes. The trace file, TRACE the request and review (or the tkprof).

Hemant K Collette

How to investigate DB performance problems?

How to investigate DB performance problems?

You can refer to the oracle support doc: Doc ID 1362329.1

In case you have difficulties to access the Oracle support, I copied some of these inline content:

K.

[ORACLE 10.2.0.5] performance problem with high enq: TX - line lock contention waiting

Hello
I have a performance problem involving high line response to locking conflicts that I can't wait.
The application updates the specific table about 1200 times for my AWR report representing 2 hours.
This update represents 98% of the db_time and wait to enq: TX - line lock conflict is:
-83 796 waiting
-245 441 seconds
The application updates a row both an unprecedented SELECT UPDATES. This is the query:
Update my_table set creationtime =: 1, modificationdate =: version 2, =: creationuserid 3, =: fortesting 4, =: 5, modificationuserid =: participation_uuid 6, =: actimenucontact 7, =: allegroid 8, =: campaign_uuid 9, =: desjardinsemployeetype 10, =: effectiveparticipationtype 11, =: family_uuid 12, =: fetchpasseport 13, =: 14, healthratingfilldate =: hoursofsleeptimecommitment 15, =: igaid 16, =: initialsubscriptiontype 17, =: minutesofsleeptimecommitment 18, =: goal 19, =: partner_uuid 20, =: 21, promotionalemails =: 22, readrules =: sccsubscriptionprovenance 23, =: = subscriptionprovenance 24,: 25, supportemails =: where the 26 uuid = : 27 and = version: 28
Clause where is filtering on uuid which is the primary key of my_table.
The uuid is a specific user and two users would not update the same row, what could be the reason for this wait event?
Thank you

Why guess when you know.

Watch ASH data for these events to the queue.

He'll tell you sql id as well.

took vcap5-dca exam today and had laboratory performance problems

Most of the time, the review ran ok. I was very impressed by the performance of the lab environment. I can say that I had to skip two questions I knew because the keyboard and mouse of entry would not work. After spending 5 minutes trying to get my keyboard and mouse works on one of the questions, I had to jump even if I knew the answer. It is unfortunate that, for a review of 400 dollar, you have performance problems that limit your ability to perform the review. Overall, I think I did pretty well. I was able to get about 19 questions answered to jump through the other 7 due to the performance and time. I acknowledge that the pace I started to was a bit slow, but laboratory performance degrade my ability to perform my best on this review. I would recommend to anyone who takes this review to make sure that you use a good test center. I drove 60 miles to my test center to make sure that I went to a good and always fell on issues. I could only imagine what it would have been like if I went to one of these centers of samples of groceries with a metropolitan WAN connection for internet :-) Now wait to see if I've spent 15 days.
Has anyone else had this kind of questions? If so, where did you register a complaint?

Hello

The delay is a common thing for everyone outside the USA. She varies the amount of delay but it has been around since the VCAP-DCA 4. In regard to complaints they should normally be done the day of the test to the exam administrator otherwise pearson cannot do much.

Good luck for the result!

Gregg

Troubleshooting performance problems with the mail server on the virtual computer

I have a virtual machine with a mail server installed on a windows OS. But all of a sudden, my clients on the server response times decreased significantly.
This could be the cause of this problem?
It could be to do with the network, storage, or resources, but I don't know what steps I could take to determine fault
Would someone mind helping me with the steps I could take to solve this problem?
Thank you

The below document page 20 and 21 accurate meters, you should look at the virtual machine level.

http://www.VMware.com/files/PDF/Exchange_2010_on_VMware_-_Best_Practices_Guide.PDF

These counters can be used for any virtual computer not only Exchange. If you are familiar with ESXTOPs I believe that the best way to deep dive in the diagnosis of performance problems. Also here is a link for these parameters.

http://KB.VMware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalID=1008205

This link takes you to http://communities.vmware.com/docs/DOC-9279 which has the description on the stats.

New performance problem

Hi all
We encounter a performance problem once again
The batch deletes lines 1 M every night which took regular 30 minutes.
But last night (midnight) it took more than 2 hours and crashes.
Help if I run gather_schena stats regularly when it is constant DELETE on the table?
Please help me check our ASHES, AWR, ADDM to solve the problem.
ADDM
https://app.box.com/s/7o734e70aa2m2zg087hf
ASH
https://app.box.com/s/xadlxfk0r5y7jvtxfsz7
AWR
https://app.box.com/s/x8ordka2gcc6ibxatvld
Thank you...
zxy

yxes2013 wrote:

Are you & twin SB

Good for say you that because you have a minimum of charges/taks in you and your company is paid too high

For me, Im so overwhelmed with a lot of assignments include not less than the ff:

GG, DBV, TDE, Ebiz, MySql, Sqlserver, Db2, Sybase, Foxpro, RMAN, Dataguard, database security audit, database administrator, etc., & etc.

And I get paid only USD1, 500 per month. Is this reasonable?

Why is - all the dba from here seem to not complain about their jobs?

Do you mean most of you have minimum assignments with your respective companies and is paid at the top?

Thank you

Fair enough. He turns off the stage (again) and is going no where.

Action mod.: blocking of the thread.

Nicolas.

calculate statistical performance problems

Similar Questions

Maybe you are looking for