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.

Tags: Database

Similar Questions

  • 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 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.

  • 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.

  • Just a quick question about the appearance of the cluster

    Hello, this is Matthew, just a quick question about the appearance of the cluster

    Is there anyway to rearrange the order of the elements? I know cutting automatic resizing, I could move the items by hand. But if I use the cluster somewhere else, I have to move the items again...

    What I want is of the order of "Mx My Mz Ax Ay Az Temp voltage CS".

    Is there a quicker way to deal with?

    THz so much!

    You can right-click the border of the cluster and choose "arrange control cluster...". "The user interface is not as intuitive as it could be, but it will allow you to reorder items. If you use the cluster in several places, you should make the cluster in a type definition and replace all current uses of the latter with the type definition, so that they all match. They will also update automatically when you make changes to the type definition. Otherwise, you could have a situation where you have multiple groups with the same data types, but with elements in a different order, and your data will be getting a new denominated, when wire you the whole clusters. The data will be in the same order, but given that the item labels are in a different order, you can't get the data you want, when ungroup you by name.

  • MAA - RAEVEN & DataGuard conceptual question about Photo (10.2 doc)

    Hello experts,

    I have question about the figure:
    'D.1.2 putting into place of a primary of multiple instances with a multi-Instance standby'
    http://download.Oracle.com/docs/CD/B19306_01/server.102/b14239/IMG/rac_arch.gif

    page http://download.oracle.com/docs/cd/B19306_01/server.102/b14239/rac_support.htm

    Detailed explanation is provideded here:
    http://download.Oracle.com/docs/CD/B19306_01/server.102/b14239/img_text/rac_arch.htm

    It is said:
    «This illustration shows a primary database archiving online redo logs to a database, multi-instance multi-instance ensures in a Real Application Clusters environment.» In this configuration, there are two instances of primary database: has the instance primary and primary Instance B. There are also two instances of sleep: standby receiving Instance C and standby recovery Instance D. The definition and purpose to receive the bodies and recovery is described in the text that follows this illustration. Each primary instance uses a LGWR to write again online newspapers and recovery logs archived local processes on the primary instance. In addition, the process LGWR on the primary Instance a sends its changes over an Oracle Net network to the RFS in First Instance B process and to the RFS process on Standby receiving Instance C. primary Instance B sends its changes over an Oracle Net network to the RFS on Standby Recovery Instance D process. The RFS process on each standby instance written in local newspapers do sleep. This figure also shows how the process ARCn on Standby receiving Instance C sends its changes over an Oracle Net network to the process on Standby Recovery Instance D RFS. The process on Standby Recovery Instance D ARCn also archives its changes in newspapers local archived redo. »

    Question I would like to ask because I'd like to better understand the Internals:
    (1) why it is written that LGWR writes Archives newspapers and not ARCH process? Is this some sort of error doc. ?
    (2) what is the reason that LGWR sends redo changes made to the FIU to the instance of the same (primary) cluster? What is the purpose? What happens if it has many nodes in the primary? This means that this instance would be multicast it to each of them in this way?
    (3) on the backup site: MRP is located on the D instance in this scenario? (1 single standby instance is applying the data but several can recive redo and write it to the SRls)?

    (1) why it is written that LGWR writes Archives newspapers and not ARCH process? Is this some sort of error doc. ?
    Yes, it seems that this picture has been simplified, and it does not show the level of appropriate detail.
    (2) what is the reason that LGWR sends redo changes made to the FIU to the instance of the same (primary) cluster? What is the purpose? What happens if it has many nodes in the primary? This means that this instance would be multicast it to each of them in this way?
    This shows what we call "cross instance archiving". If you enable this, one storing data on multiple nodes. So if you are in a cluster, and archive locally, if this node dies, so how you get that archive the data to retrieve, put the archives in several places gives you extra security for those who are paranoid. I think that it was more useful in the days where Oracle shipped just archiving logs, now that lgwr writes to the remote node, you are less likely to need it.
    (3) on the backup site: MRP is located on the D instance in this scenario? (1 single standby instance is applying the data but several can recive redo and write it to the SRls)?
    Yes one instance applies to the remake.

  • Where can I ask a handful of general order questions about Firefox OS as a consumer and get the answers? (a forum or something similar)

    I have about seven questions about Firefox OS. I'm potentially interested in opting for a tablet that works instead of Android or iOS. I couldn't find a place to ask questions like that. Thank you!

    https://support.Mozilla.org/en-us/forums/Firefox-OS-English-Forum

    There is another option.

  • About a month ago I posted a question about iMovie and not being able to "share". I solved the problem thanks, so no more emails!

    About a month ago I posted a question about iMovie and not being able to "share". I solved the problem thanks, so no more emails!

    Hi Michael,

    If you want to stop receiving notifications by electronic mail, in the thread, that you have created, then I suggest that you follow the steps below:

    One time connected to the Apple Support communities, visit your mini profile and select manage subscriptions.

    Content

    To manage this content, you are currently subscribed and changing your preferences, select the content.

    Select next to see what content you are currently following.  Note that any thread you are responding you subscribe you automatically to this thread.

    You can select to terminate a subscription to a thread.

    Learn how to manage your subscriptions

    Take care.

  • Where and how do I ask questions about Email?

    The Commission opening thingy is complicated, I have trouble finding my way around it.  I don't know where or how questions about email.  Is there an easier way than trawling through all these icons?  Like a list or something - as you get on all other forums on the internet?

    It would be a good starting point: Mac mini

    Or the appropriate under MAC Os and system software section.

    You might also find this useful: find answers and new questions

  • Many questions about the HP 50 g

    I just ordered a HP 50 g graphic calculator! I have a few questions about this so hang on tight! What is the USB connector on the calculator? Mini-USB, micro USB, or is it a connector owner, "custom". Then, it would be possible to connect a keyboard to the HP 50 g by using the USB Port? Is there a program that would make this possible? (I doubt it would work if you plug it..) I have a Mini-USB to female USB 2.0 connector. So I could just plug the keyboard right, provided that the port on the calculator is a Mini-USB Port. If this is not so that I could use my female to female USB adaptor to connect the keyboard.

    Thanks for your help,

    Zekelegge

    Hi!, zekelegge:

    Welcome to Forum!

    The only printer I use, 50G calculators, 48GII, 39GS and 17BII +, as well as many former models, such as the 48 s/SX/G/G + / GX, 38 G, 39 G, 42, 41 C/CV/CX and 28 C/S, is the infrared printer HP 82240 b. If you can use the emulator HP50G, on the PC, the simulator of the printer, you can download, since... http://www.HPCalc.org/details.php?id=7386

    The printer... HP 82240 B

    The cable series...

    The description... http://www.allenwan.com/hpcalcserialcable/

    The original cable...

  • Question about warranty after exchanges of RAM on Satellite module series

    Hello together,

    I have a question about the warranty. My laptop is 5 months old and now I want to improve it with more RAM. I want to do it myself, and now I'm wondering if the warranty will be getting hurt.
    Thanks in advance for your answers.

    Concerning

    TiG3r

    For new models of laptops the RAM upgrade described in s operating manuals. It is public act and it shows the owner of the notebook how do RAM upgrade only.

    If Toshiba offers something like this he won't have the negative influence of valid warranty.
    Buy a compatible module and RAM upgrade.

    Bye and good luck!

  • A few quick questions about the Satellite P300 - 18 M

    Hello

    I am about to order a Satellite P300 - 18 M and have just a few questions about this:

    (Although these may apply for other similar Satellite/Equium P300.)

    Is the touchpad made by / to use the Alps or Synaptics driver?

    It has the Wireless N network card, is the map made by Intel or Realtec?

    And also, someone has experience with face recognition software to connect?

    If so, does it take a lot of time to initalise?

    Thanks in advance,
    Jack

    Hello

    The P300 - 18M belongs to the PSPCCE series.
    It uses the two touchpad drivers. Synaptics & Alps Electric
    There is an Intel 802.11agn minicards WLan installed

    The face recognition is a nice tool but local to be illuminated and you must take your face the right of the post sometimes it of not very easy

    But if you face is recognized correctly that the connection should work properly

    See you soon

  • Questions about nb200

    I recently bought a mini nb200 with windows 7 and I have a few questions about it. How can you set up your email, I was willing to put up with Outlook Express, but I don't know if I can. Also, is there a way to get a paper copy of the user manual?

    You can download and install a new e-mail program when you download Windows Live via Microsoft Update. You can select the components of Windows Live, you want to include by checking a box for each you want. I installed all the. Or you can use Outlook, if you have Microsoft Office and you want to install it.

  • A few questions about the difference between the Satellite P70, L70, S70

    Hello, I have a lot of questions about the P70, L70, S70 series that come with a 1920 x 1080 panel.

    (1) what are the differences between the L70 and S70 series? With the exception of the RAM and HARD drive capacity, books seem pretty identical.

    (2) P70, L70, S70 doesn't support a 2nd HARD drive or it's just the P70 series that support?

    (3) all the three (P70, L70, S70 series) come with the same TFT panels?

    (4) of the above series, which supports mSata?

    (5) all the model of each series are delivered with support from mSata? For example, it could be that L70 - a - 13 m supports mSata is not the case of the L70-a-146?

    (6) all the foregoing, are delivered with a S - ATA II or III S - ATA interface?

    (7) who is the best of these series listed? I'm trying to understand what makes the big difference of S70 to P70 except for the envelope for example.

    Thank you in advance.

    > (1) what are the differences between the series L70 and S70? With the exception of the RAM and HARD drive capacity, books seem pretty identical.

    What models Sat L70 and S70 do you mean exactly? There are different L70-xxx-xxx and S70 models on the market that supports different hardware specifications.

    (> S70 2) P70, L70, support a 2nd drive HARD or is - it just the series P70 that support?
    As you can see in this [Sam P70 HDD replacement document, | http://aps2.toshiba-tro.de/kb0/CRU3903II0000R01.htm] the P70 series supports the 2nd drive Bay HARD, BUT even if there is a 2nd HARD drive Bay, this does not mean that you can use the 2nd HARD drive. In the case where the 2nd HARD drive Bay are equipped with HARD drive connector, you can use the 2nd HARD drive

    I also found the [Sam L70/S70 HDD replacement | http://aps2.toshiba-tro.de/kb0/CRU3703HG0000R01.htm] the document on the Toshiba page and there I see this 2nd HARD drive Bay is not available

    (> 3) all three (P70, L70, S70 series) come with the same TFT panels?
    See point 1). Different P70, L70, S70 models were equipped with different material parts.

    (> 4) of the series above, which takes in charge mSata?
    As far as I know that some P70 models are equipped with an mSATA SSD of 256 GB.

    (> 5) do all the model of each series are delivered with support mSata? For example, it could be that L70 - a - 13 m supports mSata is not the case of the L70-a-146?
    See point 4) not all models supports the same hardware specifications

    (> 6) all of the above, come with a S - ATA II or III S - ATA interface?
    I don t think that SATA III is supported. I guess it would be SATA II

    (> 7) which is the best of these series listed? I'm trying to understand what makes the big difference of S70 to P70 except for the envelope for example.
    Not easy to answer because there are too many models released in Europea.
    And not all models are available in each country. So I guess you will have to look for the models that have been released in your country.

  • Satellite L650-11R - a few questions about this

    Hello there :-)

    I have a few questions about my new L650.

    I've never had Toshiba notebook before so I apologize if some of these "quirks" are actually quite normal on this laptop. I really want to hear the views of other users L650.

    1)
    First of all, by the time I turned it on I saw gradient thin (1-2 mm) along the right edge of the screen that goes from the bottom up. Maybe "degraded" isn't the right word, but it seems that since two or three pixels are kind of discoloration or lose brightness/sharpness to the edge along the straight bezel. It is not really that visible, looks like the colors are desaturated a bit here or brightness is a bit lacking. He can be seen everywhere (bios or no color in Windows, but better in the lighter background).
    The left side (or other) seems ok, all this along the right edge is somewhat suspicious. If one can compare? I don't know if it's something to worry about or is just supposed to be like that.

    2)
    The other thing I noticed heating of the processor (Intel i3 330M) and GPU (ATi Mobility Radeon 5650). It's pretty hot here right now (I would say that the temperature is around 28 degrees Celsius when using cell phones). He is sitting on the hard surface (table) and nothing is blocking the vents.
    However, at "rest" (surf the web & regular light stuff), average CPU temp hovers around 58 ° C. It usually goes from 45 to 62 in just a few minutes after starting, and then I can hear and see (Toshiba PC Health Monitor) evacuate the kicks in. Temp goes back to 50 +, event stops and then the temp rises to 60 + until the vent is used again.
    During load full (Prime95 or some games), CPU is going all the way up to 75 degrees (the highest being 77). Fan runs at 75 to 77% during this time, never go faster.

    GPU idling is usually around 5 degrees above temp CPU (62-65). During full load (for example. Benchmark of Far Cry 2), I see a lot of stuttering with the latest drivers Catalyst and DX10 (not sure if the stuttering has something to do with heat), otherwise it is ok (no artifacts or anything like that). Temp of GPU during the loading goes up to 74-75, fan speed is locked at 30%, the frequency of the heart 5650 is 450 MHz. left side of the laptop (vent and below) is pretty hot, while the keyboard is warm (or slightly warmer who who).

    These time seems correct and safe in the long term? I'd rather events works all the time (they aren't really very strong) and keeping time less than 50, but there are just wishful pious bios control events and I can't control not the bios. :-)

    3)
    This is just General related products. :-)
    Official specifications of Toshiba L650 say ATi 5650 (between other low range GPU), but I saw a few specifications calling for there 5730 (same model 11R). Is there really a Satellite L650 with ATi 5730 GPU models?

    Sorry for the long post, I tried to include the best possible description that I could do. :-)

    Thank you in advance.

    Thanks for the additional info!

Maybe you are looking for