Question of clustering pls

Similar Questions

• A question about clusters

  Hello
  
  I have a question about the single table hash clusters. How are these paintings in a position to provide better performance? Any block can contain data from a single table, and here was the only cluster table will also create blocks with data from a table. So how or why is it more effective? I am not unclear on the concept plan and will be grateful for the info.
  
  Concerning
  
  Published by: orausern on 22 Sep, 2010 07:57

  orausern wrote:
  I did it - but the concept is what I'm looking for. The test case proves that there are benefits. but my question is why the profit returns. It's what I'm not very clearly.

  The benefit is provided
  (1) exercise you some control over the way in which the data are filled in the table
  2) most of your questions is a 'SELECT col1, col2,... '. "Col WHERE = "
  (3) the number of values unique 'col' is fairly constant
  The common approach in the situation described above is to create an index on 'col' so that applications may perform better.
  But even with the indexes in place, oracle will need to access the blocks of two segments, index and table, in order to process the requests (of the type described in step (2) above).
  When you create the table in a hash TABLE UNIQUE cluster, you can get the grouping of the data in the table itself, without the need of an index. Oracle will have access to the only segment (cluster) blocks to process requests (of the type described in step (2) above).
  The result is much less number of consistent gets for queries of the type described in step (2) above, resulting in better use of the buffer cache, which improves performance.
  I hope this helps.

• A Question abount Clusters

  Hi gurus!
  
  I have a question regarding the use of clusters. I read the following in the onlinedocumentation of Oracle:
  >
  Clusters are a method of storage of the data in table optional. A cluster is a group of tables that share the same data blocks because they share common columns and are often used together. For example, the employees and the departments table share the department_id column. When bring together you employees and departments, Oracle tables physically stored all lines for each Department in the employees and departments tables in the same blocks of data.
  >
  After reading that I thought that the use of clusters with tables often gathered brings a benefit performance. So I made the following testscenario:

  CREATE CLUSTER cl_emp_dept (department_id NUMBER(4));
  
  CREATE INDEX cl_emp_dept_idx
  ON CLUSTER cl_emp_dept;
  
  CREATE TABLE emp
  CLUSTER cl_emp_dept (department_id)
  AS
    SELECT * FROM employees;
  
  CREATE TABLE dept
  CLUSTER cl_emp_dept (department_id)
  AS
    SELECT * FROM departments;

  After the creation of the cluster with the two tables, I tried the following:

  DELETE FROM plan_table;
  
  EXPLAIN PLAN
  SET statement_id = 'NO_CLUSTER'
  FOR
    SELECT *
    FROM   employees e, departments d
    WHERE  e.department_id = d.department_id;
  
  EXPLAIN PLAN
  SET statement_id = 'CLUSTER'
  FOR
    SELECT *
    FROM   emp e, dept d
    WHERE  e.department_id = d.department_id;
  
  SELECT     LPAD(' ', 2 * level) || operation ||' ' || options || ' ' ||
             object_name  || ' ' || cost || ' ' || cpu_cost || ' ' || 
             io_cost AS query_plan
  FROM       plan_table
  WHERE      statement_id IN ('NO_CLUSTER','CLUSTER')
  CONNECT BY prior id = PARENT_ID 
      AND     statement_id IN ('NO_CLUSTER','CLUSTER')
  START WITH id = 1;
  
  --NO_CLUSTER
  QUERY_PLAN
  --------------------------------------------------------------------------------
    NESTED LOOPS   4 969871 4
      TABLE ACCESS FULL EMPLOYEES 3 73057 3
      TABLE ACCESS BY INDEX ROWID DEPARTMENTS 1 8381 1
        INDEX UNIQUE SCAN DEPT_ID_PK 0 1050 0
  
  --CLUSTER
  QUERY_PLAN
  --------------------------------------------------------------------------------
      TABLE ACCESS FULL DEPT 8 205070 8
      TABLE ACCESS FULL EMP 8 236850 8
    HASH JOIN   17 3979365 16
      TABLE ACCESS FULL DEPT 8 205070 8
      TABLE ACCESS FULL EMP 8 236850 8

  So my question is: "why are the costs, the cpu_costs and the much higher io_costs when I use the cluster. I well using the cluster is a performance advantage not a disadvantage. Can someone please explain it to me.
  
  I look forward to your responses and I thank you for all your efforts to assist me.
  
  Yours sincerely
  
  Florian W.

  As mentioned above - your tables are really too small to make a reasonable comparison. But you have also exaggerated the differences by allowing the cluster size by default for the entire block, so your data in cluster probably take more space than your data nonclustered. (The choice of the best cluster size is often the big problem with the use of clusters, incidentally).

  It is preferable to use dbms_xplan to get execution plan output - your example provided for the cluster looks very strange.
  Example of syntax for your calls "explain plan":

  select * from table(dbms_xplan.display(null,'NO_CLUSTER'));
  select * from table(dbms_xplan.display(null,'CLUSTER'));
  

  The first parameter is the name of the table used as the plan table (default PLAN_TABLE when left null), the second is the value of the statement_id that report.

  Concerning
  Jonathan Lewis
  http://jonathanlewis.WordPress.com
  http://www.jlcomp.demon.co.UK

  "For every expert there is an equal and opposite expert."
  Clarke

• Question about Clusters and host bus adapters

  I have worked with ESX in a non-clustered environment, but I'm looking to experiment with clusters.

  If I have two host servers in a Vmware HA cluster (using VC 2.5)

  And I have these two server hooked up to an FC SAN, each with a FC HBA.

  Now, I create two virtual computers that will be part of a Windows / SQL cluster, do I need a second set of FC HBAS in each server?

  Now if I add a second series of cluster of virtual machines, this time window / cluster exchange, I have to add yet another set of HBAs?

  I understand that if I had a physical server, I need an HBA in each... but unsuire how sharing equipment works in vmware. It's the virtual FC HBA for ESX?

  Thank you!

  bilbus wrote:

  OK thanks, so as to be clear...

  so, if I have one HBA for each ESX cluster server (assoming I don't need double tracks) that's all I need?

  OK - all you need is 1 HBA per server.

  Once the server has a FC HBA (even if it is used for the ESX cluster server it self) I now have access to a virtual FC HBA on the set of my virtual machines?

  No, this is not how ESX.  Each VM has a virtual card SCSI attached to a VMDK on your data store, which is accessed throught he FC HBA above.

  Is there a rule on how many servers can share an HBA (another bandwidth then)?

  There may be, but its mainly led by IOPS / s and bandwidth.

  -Matt

• Question of clustering

  Hi all
  
  I have correctly installed and configured the SOA suite 11.1.1.3 & BAM in a wls 10.3.3 in a UNIX domain. Now when I try to create the SOA and BAM (name: soa_Cluster & bam_cluster) with two managed servers: bam_server2 & soa_server2 on another machine, I get these errors when starting servers managed to soa_Cluster and bam_cluster and managed servers, put into service in a State of the ADMIN: -.
  
  
  < 29 November 2010 12:20:59 PM CST > < error > < hats > < BEA-149205 > < failed to initialize the application "usermessagingserver" due to the error weblogic.application.ModuleException: cannot load the webapp: ' / sdpmessaging/userprefs-ui.
  weblogic.application.ModuleException: cannot load the webapp: ' / sdpmessaging/userprefs-ui.
  at weblogic.servlet.internal.WebAppModule.prepare(WebAppModule.java:404)
  at weblogic.application.internal.flow.ScopedModuleDriver.prepare(ScopedModuleDriver.java:176)
  at weblogic.application.internal.flow.ModuleListenerInvoker.prepare(ModuleListenerInvoker.java:199)
  to weblogic.application.internal.flow.DeploymentCallbackFlow$ 1.next(DeploymentCallbackFlow.java:507)
  at weblogic.application.utils.StateMachineDriver.nextState(StateMachineDriver.java:41)
  Truncated. check the log file full stacktrace
  
  Caused by: weblogic.management.DeploymentException: [HTTP Session: 100083] the webapp: / sdpmessaging/userprefs-ui in the application: usermessagingserver has the value PersistenceStoreType: replicated_if_clustered for http sessions, but the target list contains all members of the cluster: bam_cluster. Inhomogeneous deployment for replicated sessions is not allowed.
  at weblogic.servlet.internal.session.TargetValidator.validateTargetting(TargetValidator.java:31)
  at weblogic.servlet.internal.session.SessionContext.getInstance(SessionContext.java:379)
  at weblogic.servlet.internal.WebAppServletContext.initSessionContext(WebAppServletContext.java:1233)
  to weblogic.servlet.internal.WebAppServletContext. < init > (WebAppServletContext.java:443)
  to weblogic.servlet.internal.WebAppServletContext. < init > (WebAppServletContext.java:484)
  
  
  Just trying to create a cluster with 2 managed servers still say (in a single box of unix bam_server1) & bam_server2 in another unix machine, but this error indicates clearly that "usermessagingserver has the value PersistenceStoreType: replicated_if_clustered ', according to me, there is an error I have while clustering. And I maybe need to retarget to something else... someone can you please help me solve this problem.
  
  Thank you
  Sam

  Look at the error message:

  The webapp: / sdpmessaging/userprefs-ui in the application: usermessagingserver has the value PersistenceStoreType: replicated_if_clustered for http sessions, but the target list contains all members of the cluster: bam_cluster. Inhomogeneous deployment for replicated sessions is not allowed.

  I'd say it's pretty descriptive. When there is replicated_if_clustered of an application setting, you will need to deploy the application on all nodes in the cluster. Just make sure that your nodes are running and target your application across the cluster (instead of individual nodes).

• HP pavilion 1410n starts and questions of fan pls help

  Hi everyone im new and here I need help with my HP Pavilion 1410n. Ive had my office stored for like a year since I have a laptop I don't use any, yesterday, I decided to take it to find some of the files I needed, but when I turned it on it does not start at all, and all this I can hear is the fan connected to the motherboard running non-stop and does not show me anything in the monitor.ive never had this kind of problem before Ive had for about 6-7 years ago until now since I bought it thanks in advance

  Try a hard reset as follows:

  (1) remove the battery and unplug the DC power adapter.
  (2) press and hold the power button for more than 15 seconds.
  (3) plug the DC power adapter (leave the battery for now)
  4) press the power button

  Please post here your positives or negatives in your thread.

• question about clustering configuration in LCDS3.1

  In the doc LCDS3.1, it is said "using the"
  shared-backend = "false".
  the configuration attribute does not have a single, consistent view of the data in the cluster. When two clients are connected to different servers in the cluster, if they make conflicting updates to the same data value, each server applies these changes in a different order. The result is that clients connected to a server to see a view of the data, while the clients connected to other servers have a different view. The conflict detection mechanism does not detect these changes as long as conflicts. If you need a single view of data where both clients are updating the same data values at roughly the same time, use a database or some other mechanism to ensure that the regular blocking is done when updating the data in the back-end system".

  This means using the static-backend = "true" IS guaranteed a single consistent view of the data in the cluster? Or it means that I must use the level of database locking in both cases?

  Here is some info on this to Mete Atamel, engineer of the LiveCycle Data Services team:

  When shared-backend = true, a single consistent data display is guaranteed, but it's not because of the shared-backend itself, but rather the fact that once shared-backend = true, there is a single database used by multiple instances of the LCD screens.

  Imagine a scenario where you have two LCDs nodes in a cluster with the destinations of two data, on node1 and Node2, on management and they are grouped (via JGroups). You can have these destinations maintain their own DB, and when a change occurs in a node¹s destination, destination of the other node gets a message JGroups and updates its own DB. It works but it is not ideal because at some point, you¹re try to maintain the State of cluster even on two different DBs and there are times where connected clients do not get a consistent view of the data.

  Instead, you can use a single database and have the two destinations, talk with the same PB, if you have a single consistent view of the data. When possible, you enable property shared-backend in the destinations of LCD screens, in this way when a single destination updates of the DB and send a message of JGroups the other destination, the other destination do not have to update the DB again, instead she simply passes the update message to clients connected without sending it to the adapter.

• Number of Clusters

  We are mounted to finally put VC2.5 in vSphere and I have a question about clusters in vSpere 4 update1. At this moment we have tough 6 hosts (nodes) running s 70vm each. As we look at adding additional nodes xseries or blades, should we be creating additional clusters or just keep everything in 1 cluster. Are there disadvantages of pro with who (stipulated the rate cap for vSphere cluster 8, 9 or more nodes etc.). I mean I guess my question is why create a 2nd cluster just because I add more hosts? A colleague suggests an additional cluster will alleviate cpu contention, but by adding more nodes to an existing cluster should do the same thing in my mind. Thank you for advice or links to best practices.

  vmdawg

  The most important thing that could lead you to a new grouping would be CPU compatibility - you have CVS available, but if it is not currently enabled will require your virtual machines stop and restart them once active CVS.

  IMHO, I prefer smaller larger clusters that's easier to manage them and their virtual machines

  If you find this or any other answer useful please consider awarding points marking the answer correct or useful

• Pls suggest me a good model for a decimal field

  Hello

  Pls help me by suggesting a good model for a decimal field to the value of the amount, my problem is as below

  The lengths are as below,

  

  I specified the reason for this field as below (its as-> {} null | zero {} | num {($z9,999.99)} | {} num.decimal), pls note that I did ' t put any model in the tab CHANGE

  

  (1) but, when the user enters the value get as on the form.

  

  But I want that the foregoing as just $1.10 or $1,1

  (2) in addition, make me as below for any other value,

  

  But I want that the foregoing as just $0.60 or $0,6

  (3) another case like below,

  

  But I want him above looks like just $12 345 (if possible, if this is not possible it is OK)

  (4) when the form is rendered on the browser see the field as below,

  

  But, once the user put the cursor/clickedon field zero is failing as below,

  

  but, I don't want default zero as above, with just white is eager (if its possible so this is nice, if this is not possible it is OK), in fact, I have another decimal field as below with the model, (pls note this field also has a ppatern of CHANGE as well)

  

  The EDITING model is as below,

  

  so this aspect of the rendering of form field is as below,

  

  Once the user put the cursor / clikced bile field looks like as below

  

  It's fine for me instead of zero defect as described in question # 4

  Pls suggest me a good role model for my 4 questions above

  Thank you

  So use the locale of the system of the spectator and just put the model as:

  NUM. Currency {ZZ, zz9. ZZ}

• Buy a lens that is suitable for portraits and photography macro?

  I have a canon 600 d.

  I want to buy a lens that can work as well for portraits and also for the macro. Is this possible?

  Budget is about $400. Stretch Max $ 600

  I was recommended 50 mm f1.8 for portraits.

  It will also work as a macro lens?

  Thanks you.what you hear by 100 mm too long for the portrait? a long time in the size of the lens? What's really important.
  I'm sorry if you find my questions too basic pls.

• ACS version 3.3

  Hi, in our environment, we have Cisco ACS v3.3 in windows 2003 and try to improve v4.1.4 ACS. but that v3.3 on v4.1.4 data replication is originally a question.

  Pls let know us is there a way to make the replication of data with this different code.

  Thank you

  Gopinath V

  Hi Gopinath,

  For the replication process, primary & secondary servers should be in the same version.

  Please upgrade primary & secondary to 4.1.4 and initiate replication.

  Excerpts from the User Guide:

  "All of the SACRED that is involved in replication must run the same version of the ACS software. For example,.

  If the ACS primary runs ACS version 3.2, all secondary ACSS should work ACS version 3.2.

  Because patches can introduce significant changes to the internal database of ACS, we strongly

  "recommend that ACSS involved in replication use the same patch level.

  If the two ACS (primary & secondary) are in the same version, and still you are facing some questions, let me know.

  Thank you

  Séverine

• Impossible to import metadata using 11 g OBI administration tool

  Hello people,
  I'm new to OBI. I am trying to build a RPD. To do this, I try to import database tables, but I can't. I created DSN and also tried using OIC but no help. OBI the Admin client tool and database is installed under the account of the other user. Is this the reason why I'm unable to import tables?
  I get this error:
  Error internal-failed to initialize the NLS during loading driver - when I try to use ODBC &
  Connection failed - when I try to use OCI.
  
  Can someone shed light on this question.
  
  Thank you
  
  Published by: 950330 on January 21, 2013 11:17

  Use the calling Interface: by default (OIC 10 g / 11g)
  Name of source data-> hostname: port/ServiceName or name TNS
  as
  localhost:1521 / XE

  Make sure that you can ping connect to the database using tnsentry or
  tnsping
  from the command prompt.

  Let me know for questions if brand pls help

  Published by: Srini VIEREN on January 21, 2013 13:17

• How do I check how much time running on the query server

  Hello
  
  I work as a junior DBA. The guys on the team application gives the SID from v$ session and asking me how long running this query.
  I'm connecting ther server and get the PID and cpu utilization on unixbox, but cannot get it question time.
  Pls help help get duration/how running queries. Verion Oracle's 9i.
  
  SQL > select * from v version $;
  
  BANNER
  ----------------------------------------------------------------
  Oracle9i Enterprise Edition Release 9.2.0.8.0 - Production
  PL/SQL Release 9.2.0.8.0 - Production
  CORE Production 9.2.0.8.0
  AMT for Solaris: 9.2.0.8.0 - Production Version
  NLSRTL Version 9.2.0.8.0 - Production
  
  Kind regards
  Vincent

  >
  This means 113547 minutes. Right?
  >

  Yes, it's true. But it is not a few Minutes but microseconds. ELAPSED_TIME is microseconds (10g or higher. Not sure about 9i. Please check the doc)

  >
  Number of executions increasing very fast now its 29999. with in wrigting mail its increase. to get the output why this query execution of this many times.
  >

  This means there are a few sessions by running the same query (which is fine). Only, you will need to tell us why the query is run that many times :) Maybe, you need it!

  Published by: user12035575 on September 11, 2011 14:38

• I use windwos xpsp3 teacher, my question is wn I install winxp after 6 months my hardware (motherboard) will be too bad so I'll be changed motherboard so I can use my OS still this material, I've heard could not appu .so wy use winxp.so me.pls help

  I use microsoft Windows xpsp3 teacher that my question is wn I install winxp after 6 months that my hardware (motherboard) will be too bad so I'll be changed motherboard so I can use my OS still this material, I've heard could not appu .so wy use winxp.so me.pls help

  You can reuse your current XP only if the version is not OEM.  OEM is a version that must be supported directly by the manufacturer of the computer first.  OEM is always linked to the first PC, the license is activated on.  This license is suppused to 'die' with the PC.  Now, there is a "grey area" on the definition of a PC, but accepted the more common definition all came to the conclusion that the 'mother' is the building block for the definition.  Therefore, change of motherboards would mean you have a "new" PC

• Questions about the implementation of clusters of storage and DTS on active cluster

  I have a few questions about the implementation of clusters of data store and SRS storage in VSphere 5.1.

  We have a data center with about 15 HP blades and a few servers non-lame.  The hosts are all either VSphere 5.0 or 5.1.  Our back-end storage is an EMC VNX 5700 with about 20 stores of data.  All the VMWare is managed by VCenter 5.1 running on a dedicated physical machine.

  Currently, each data store is used individually; No piles of data store are put in place.  When a new virtual machine is created, the administrator usually chooses the data with the most space store.  Periodically, we will go back and manually storage vMotion machines to balance the load.

  Recently, we have expanded the VNX storage, so we now have LUNS in different pools, with different levels of performance.

  What I would do is set up for the data store clusters, so that us when a virtual machine is created, the administrator must not know what data store is in which pool.  Also take advantage of the DRS storage so that the burden of storage will be "refine" himself a little.

  I know the setting up of a cluster of storage in a 'clean' environment is quite simple, but my concern is creation/conversion data warehouses existing in active production in the cluster, while they are used.

  If I access the screen "data warehouses and store data from clusters ', right-click on the data center and create a cluster, and then move the various data stores in the new cluster, will there be an interruption in the production running systems?

  I also wonder about activation of storage i/o control.  It is not currently enabled on data warehouses.  I know it's useful for the StorageDRS, but will be allowing any negative impact on the system?  If I turn on the warehouses of data, is there anything else I should do or set in addition to all that allows him?

  Finally, we are in the process of Site Recovery Manager configuration.  SRS has an impact on the configuration of the data clusters and DTS store?

  Thanks in advance for your comments.

  Mike O.

  Gregg Robertson wrote:

  Hello

  Clusters of data and using DTS store doesn't impact RS even if you create a cluster of data store for storage of replicated data and another for everyone else, this way you machines virtual you want to replicate are not moved on warehouses of data that are not replicated , but will still have the ability to move if there is a conflict.

  Gregg

  With all due respect but SRM and DTS don't go together. SRM does not at all support the use of DTS. So if you are configuring SRM forget DTS for now.

  Reason for this is that SRM knows no DTS and VMs can be in flight when a failover should occur, and bad things can happen. Also, the protection breaks when a virtual computer is moved between data warehouses in a cluster data store.

  In short: don't go there.