Search Results

Search found 3295 results on 132 pages for 'solaris cluster'.

Page 86/132 | < Previous Page | 82 83 84 85 86 87 88 89 90 91 92 93  | Next Page >

  • Multicast Prblem in Linux Red Hat server

    - by Uma Kant Verma
    I am facing a very strange behavior on my server. I am using a linux cluster.On fail over server my broadcast application halts after 5 minutes.But it is working fine on Primary server. My application process listening on port and send broadcast.It is working 5 mins after that it wait on socket for further packets but it didn't receive anything. Same process is working fine at Primary server. Plz help me.

    Read the article

  • percentage of memory used used by a process

    - by benjamin button
    percentage of memory used used by a process. normally prstat -J will give the memory of process image and RSS(resident set size) etc. how do i knowlist of processes with percentage of memory is used by a each process. i am working on solaris unix. addintionally ,what are the regular commands that you use for monitoring processes,performences of processes that might be very useful to all!

    Read the article

  • Benefits to private networks between virtual machines on an ESXi host?

    - by arex1337
    I'm planning this development environment with a few database servers, and originally thought I would have a few private networks. I then thought it might be unnecessary as the ESXi cluster already provides redundancy with 4 NICs (in my case) and should manage the network traffic pretty intelligently, right? Two private networks Zero private networks What are the advantages/disadvantages between the two shown configurations - on an ESXi 4.1 host?

    Read the article

  • C++: How to count all instantiated objects at runtime?

    - by nina
    I have a large framework consisting of many C++ classes. Is there a way using any tools at runtime, to trace all the C++ objects that are being constructed and currently exist? For example, at a certain time t1, perhaps the application has objects A1, A2 and B3, but at time t2, it has A1, A4, C2 and so on? This is a cross platform framework but I'm familiar with working in Linux, Solaris and (possibly) Mac OS X.

    Read the article

  • bash split text into limited character buckets (array member)

    - by soField
    i have text such as http://pastebin.com/H8zTbG54 we can say this text is set of rules splitted by "OR" at the end of lines i need to put set of lines(rules) into buckets (bash array members) but i have character limit for each array member which is 1024 so each array member should contain set of rules but character count for each array member can not exceed 1024 can anybody help me to do that solaris 10

    Read the article

  • problem installing netbeans on freeBSD

    - by solomon
    hi guys, I want to install netbeans on my freeBSD7.2 machine.I followed the instructions listed in http://typo.submonkey.net/articles/2007/7/13/netbeans-on-freebsd But, when i try the third step, which is running the netbeans using "./netbeans/bin/netbeans" i found the following error "etc/netbeans.cluster" file cann't be found. What do u think could be the reason. I can't figure it out, plz help me. Thanx in advance

    Read the article

  • version of ldap installed - one liner

    - by PJ
    HI, I am using LDAP which is installed in a solaris machine. To check the version of LDAP i go to /ldap and check the version installed as if it is version 5 then there is a directory of the name v5.0 and so on. After getting into the directory i check the directory structure. Can anybody tell me is there any shortest way or one liner to check the version of ldap installed in my machine. Thanks in advance.

    Read the article

  • What "Drained" mean in this llstatus result?

    - by xslittlegrass
    I'm trying to run jobs on the cluster of our university, and I come across this result when I trying to see the status of the cluster: llstatus Name Schedd InQ Act Startd Run LdAvg Idle Arch OpSys pandora001 Down 0 0 Run 29 41.04 9999 POWER7 AIX61 pandora002 Down 0 0 Busy 32 32.06 9999 POWER7 AIX61 pandora003 Down 0 0 Drned 0 0.03 9999 POWER7 AIX61 pandora004 Down 0 0 Busy 32 32.07 9999 POWER7 AIX61 pandora005 Down 0 0 Busy 32 32.02 9999 POWER7 AIX61 pandora006 Down 0 0 Busy 32 34.01 9999 POWER7 AIX61 pandora007 Down 0 0 Busy 32 32.02 9999 POWER7 AIX61 pandora008 Down 0 0 Drned 0 0.00 9999 POWER7 AIX61 pandora1 Avail 15 5 Idle 0 0.00 86 POWER7 AIX61 llstatus -R Machine Consumable Resource(Available, Total) ------------------------------ ------------------------------------------------- pandora001 ConsumableCpus< 28-31 >< 0-31 > ConsumableMemory(28.297 gb,124.000 gb)+ pandora002 ConsumableCpus< >< 0-31 > ConsumableMemory(14.625 gb,124.000 gb)+ pandora003 ConsumableCpus< 0-31 >< 0-31 > ConsumableMemory(124.000 gb,124.000 gb)+ pandora004 ConsumableCpus< >< 0-31 > ConsumableMemory(28.000 gb,124.000 gb)+ pandora005 ConsumableCpus< >< 0-31 > ConsumableMemory(28.000 gb,124.000 gb)+ pandora006 ConsumableCpus< >< 0-31 > ConsumableMemory(14.625 gb,124.000 gb)+ pandora007 ConsumableCpus< >< 0-31 > ConsumableMemory(5.250 gb,124.000 gb)+ pandora008 ConsumableCpus< 0-31 >< 0-31 > ConsumableMemory(124.000 gb,124.000 gb)+ pandora1 ConsumableCpus< 0-7 >< 0-7 > ConsumableMemory(32.000 gb,32.000 gb) It seems that the pandora003 and pandora008 are in the status of "Drned". What does that mean? Why I can't use the resources of at thoes nodes? Thanks.

    Read the article

  • Moving Sharepoint 2003 databases to a new SQL Server

    - by GardenMWM
    Recently we completed a new MSSQL cluster that we are planning on migrating all existing SQL databases to. This includes our Sharepoint databases, while looking into moving the databases I found the Microsoft documentation for moving Sharepoint 2007, however have not been able to find anything similar for Sharepoint 2003. Can anyone point me to a guide for moving the databases, or provide some tips, instructions or warnings? Thanks

    Read the article

  • How many Tomcat instances can one application server handle?

    - by NetworkUser
    My network engineer states I am part of a cluster where two apache application servers(512G RAM each) have 142 instances each of Tomcat (of which my company represents 6 each with 2G RAM). This seems like a lot and my latency issues move with the hour of the day - 7AM CST software functions fine, 10AM CST - system slows significantly this slowness continues until 6PM CST. My question is how many Tomcat instances can one application server handle?

    Read the article

  • How to quickly remove hundreds of thousands of files? [closed]

    - by Nick
    Possible Duplicate: Doing an rm -rf on a massive directory tree takes hours I'm running a simulation program on a computing cluster (Scientific Linux) that generates hundreds of thousands of atomic coordinate files. But I'm having a problem deleting the files because rm -rf never completes and neither does find . -name * | xargs r Isn't there a way to just unlink this directory from the directory tree? The storage unit is used by hundreds of other people, so reformatting is not an option. Thanks

    Read the article

  • How to find server is down,

    - by user363638
    I have client and server in unix machine, how can i find out my server is down and reconnect to server. without bringing my client down. I am doing connect (), but read() isiving me error Transport endpoint is already connected. I am using C++ in solaris. And what return value of read and connect should i check.

    Read the article

  • dealing with a large flat data files with a very big record length

    - by gsp
    I have a large data file that is creatd from a shell script. Next script processes it by sorting and reading several times, that takes more than 14 hours, it is not viable. I want to replace this long running script with a program, probably in JAVA, C, or COBOl, that can run on windows or on sun solaris. I have to read a group of records everytime, sort and process and write to the output sort file and at the same time insert into db2/sql tables.

    Read the article

  • SQL Server Redirection

    - by MrTehee
    We are switching from a SQL cluster to a mirrored solution. The problem is that we have a bunch of programs that would have to switch connection strings to handle the failover. Is there any way the we can set up a redirect or proxy that would take any legacy requests and forward them to the mirrored solution?

    Read the article

  • Can't re-mount existing RAID10 on Ubuntu

    - by Zoran
    I saw similar questions, but didn't find what solution to my problem. After power-cut, one of RAID10 (4 disks were) appears to be malfunctioning. I make tha array active one, but can not mount it. Always the same error: mount: you must specify the filesystem type So, here is what I have when type mdadm --detail /dev/md0 /dev/md0: Version : 00.90.03 Creation Time : Tue Sep 1 11:00:40 2009 Raid Level : raid10 Array Size : 1465148928 (1397.27 GiB 1500.31 GB) Used Dev Size : 732574464 (698.64 GiB 750.16 GB) Raid Devices : 4 Total Devices : 3 Preferred Minor : 0 Persistence : Superblock is persistent Update Time : Mon Jun 11 09:54:27 2012 State : clean, degraded Active Devices : 3 Working Devices : 3 Failed Devices : 0 Spare Devices : 0 Layout : near=2, far=1 Chunk Size : 64K UUID : 1a02e789:c34377a1:2e29483d:f114274d Events : 0.166 Number Major Minor RaidDevice State 0 8 16 0 active sync /dev/sdb 1 0 0 1 removed 2 8 48 2 active sync /dev/sdd 3 8 64 3 active sync /dev/sde At the /etc/mdadm/mdadm.conf I have by default, scan all partitions (/proc/partitions) for MD superblocks. alternatively, specify devices to scan, using wildcards if desired. DEVICE partitions auto-create devices with Debian standard permissions CREATE owner=root group=disk mode=0660 auto=yes automatically tag new arrays as belonging to the local system HOMEHOST <system> instruct the monitoring daemon where to send mail alerts MAILADDR root definitions of existing MD arrays ARRAY /dev/md0 level=raid10 num-devices=4 UUID=1a02e789:c34377a1:2e29483d:f114274d ARRAY /dev/md1 level=raid1 num-devices=2 UUID=9b592be7:c6a2052f:2e29483d:f114274d This file was auto-generated... So, my question is, how can I mount md0 array (md1 has been mounted without problem) in order to preserve existing data? One more thing, fdisk -l command gives the following result: Disk /dev/sdb: 750.1 GB, 750156374016 bytes 255 heads, 63 sectors/track, 91201 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0x660a6799 Device Boot Start End Blocks Id System /dev/sdb1 * 1 88217 708603021 83 Linux /dev/sdb2 88218 91201 23968980 5 Extended /dev/sdb5 88218 91201 23968948+ 82 Linux swap / Solaris Disk /dev/sdc: 750.1 GB, 750156374016 bytes 255 heads, 63 sectors/track, 91201 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0x0008f8ae Device Boot Start End Blocks Id System /dev/sdc1 1 88217 708603021 83 Linux /dev/sdc2 88218 91201 23968980 5 Extended /dev/sdc5 88218 91201 23968948+ 82 Linux swap / Solaris Disk /dev/sdd: 750.1 GB, 750156374016 bytes 255 heads, 63 sectors/track, 91201 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0x4be1abdb Device Boot Start End Blocks Id System Disk /dev/sde: 750.1 GB, 750156374016 bytes 255 heads, 63 sectors/track, 91201 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0xa4d5632e Device Boot Start End Blocks Id System Disk /dev/sdf: 750.1 GB, 750156374016 bytes 255 heads, 63 sectors/track, 91201 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0xdacb141c Device Boot Start End Blocks Id System Disk /dev/sdg: 750.1 GB, 750156374016 bytes 255 heads, 63 sectors/track, 91201 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0xdacb141c Device Boot Start End Blocks Id System Disk /dev/md1: 750.1 GB, 750156251136 bytes 2 heads, 4 sectors/track, 183143616 cylinders Units = cylinders of 8 * 512 = 4096 bytes Disk identifier: 0xdacb141c Device Boot Start End Blocks Id System Warning: ignoring extra data in partition table 5 Warning: ignoring extra data in partition table 5 Warning: ignoring extra data in partition table 5 Warning: invalid flag 0x7b6e of partition table 5 will be corrected by w(rite) Disk /dev/md0: 1500.3 GB, 1500312502272 bytes 255 heads, 63 sectors/track, 182402 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0x660a6799 Device Boot Start End Blocks Id System /dev/md0p1 * 1 88217 708603021 83 Linux /dev/md0p2 88218 91201 23968980 5 Extended /dev/md0p5 ? 121767 155317 269488144 20 Unknown And one more thing. When using mdadm --examine command, here ise result: mdadm -v --examine --scan /dev/sdb /dev/sdc /dev/sdd /dev/sde /dev/sdf /dev/sd ARRAY /dev/md1 level=raid1 num-devices=2 UUID=9b592be7:c6a2052f:2e29483d:f114274d devices=/dev/sdf ARRAY /dev/md0 level=raid10 num-devices=4 UUID=1a02e789:c34377a1:2e29483d:f114274d devices=/dev/sdb,/dev/sdc,/dev/sdd,/dev/sde md0 has 3 devices which are active. Can someone instruct me how to solve this issue? If it is possible, I would like not to removing faulty HDD. Please advise

    Read the article

  • Issue 15: The Benefits of Oracle Exastack

    - by rituchhibber
         SOLUTIONS FOCUS The Benefits of Oracle Exastack Paul ThompsonDirector, Alliances and Solutions Partner ProgramsOracle EMEA Alliances & Channels RESOURCES -- Oracle PartnerNetwork (OPN) Oracle Exastack Program Oracle Exastack Ready Oracle Exastack Optimized Oracle Exastack Labs and Enablement Resources Oracle Exastack Labs Video Tour SUBSCRIBE FEEDBACK PREVIOUS ISSUES Exastack is a revolutionary programme supporting Oracle independent software vendor partners across the entire Oracle technology stack. Oracle's core strategy is to engineer software and hardware together, and our ISV strategy is the same. At Oracle we design engineered systems that are pre-integrated to reduce the cost and complexity of IT infrastructures while increasing productivity and performance. Oracle innovates and optimises performance at every layer of the stack to simplify business operations, drive down costs and accelerate business innovation. Our engineered systems are optimised to achieve enterprise performance levels that are unmatched in the industry. Faster time to production is achieved by implementing pre-engineered and pre-assembled hardware and software bundles. Our strategy of delivering a single-vendor stack simplifies and reduces costs associated with purchasing, deploying, and supporting IT environments for our customers and partners. In parallel to this core engineered systems strategy, the Oracle Exastack Program enables our Oracle ISV partners to leverage a scalable, integrated infrastructure that delivers their applications tuned, tested and optimised for high-performance. Specifically, the Oracle Exastack Program helps ISVs run their solutions on the Oracle Exadata Database Machine, Oracle Exalogic Elastic Cloud, and Oracle SPARC SuperCluster T4-4 - integrated systems products in which the software and hardware are engineered to work together. These products provide OPN members with a lower cost and high performance infrastructure for database and application workloads across on-premise and cloud based environments. Ready and Optimized Oracle Partners can now leverage our new Oracle Exastack Program to become Oracle Exastack Ready and Oracle Exastack Optimized. Partners can achieve Oracle Exastack Ready status through their support for Oracle Solaris, Oracle Linux, Oracle VM, Oracle Database, Oracle WebLogic Server, Oracle Exadata Database Machine, Oracle Exalogic Elastic Cloud, and Oracle SPARC SuperCluster T4-4. By doing this, partners can demonstrate to their customers that their applications are available on the latest major releases of these products. The Oracle Exastack Ready programme helps customers readily differentiate Oracle partners from lesser software developers, and identify applications that support Oracle engineered systems. Achieving Oracle Exastack Optimized status demonstrates that an OPN member has proven itself against goals for performance and scalability on Oracle integrated systems. This status enables end customers to readily identify Oracle partners that have tested and tuned their solutions for optimum performance on an Oracle Exadata Database Machine, Oracle Exalogic Elastic Cloud, and Oracle SPARC SuperCluster T4-4. These ISVs can display the Oracle Exadata Optimized, Oracle Exalogic Optimized or Oracle SPARC SuperCluster Optimized logos on websites and on all their collateral to show that they have tested and tuned their application for optimum performance. Deliver higher value to customers Oracle's investment in engineered systems enables ISV partners to deliver higher value to customer business processes. New innovations are enabled through extreme performance unachievable through traditional best-of-breed multi-vendor server/software approaches. Core product requirements can be launched faster, enabling ISVs to focus research and development investment on core competencies in order to bring value to market as quickly as possible. Through Exastack, partners no longer have to worry about the underlying product stack, which allows greater focus on the development of intellectual property above the stack. Partners are not burdened by platform issues and can concentrate simply on furthering their applications. The advantage to end customers is that partners can focus all efforts on business functionality, rather than bullet-proofing underlying technologies, and so will inevitably deliver application updates faster. Exastack provides ISVs with a number of flexible deployment options, such as on-premise or Cloud, while maintaining one single code base for applications regardless of customer deployment preference. Customers buying their solutions from Exastack ISVs can therefore be confident in deploying on their own networks, on private clouds or into a public cloud. The underlying platform will support all conceivable deployments, enabling a focus on the ISV's application itself that wouldn't be possible with other vendor partners. It stands to reason that Exastack accelerates time to value as well as lowering implementation costs all round. There is a big competitive advantage in partners being able to offer customers an optimised, pre-configured solution rather than an assortment of components and a suggested fit. Once a customer has decided to buy an Oracle Exastack Ready or Optimized partner solution, it will be up and running without any need for the customer to conduct testing of its own. Operational costs and complexity are also reduced, thanks to streamlined customer support through standardised configurations and pro-active monitoring. 'Engineered to Work Together' is a significant statement of Oracle strategy. It guarantees smoother deployment of a single vendor solution, clear ownership with no finger-pointing and the peace of mind of the Oracle Support Centre underpinning the entire product stack. Next steps Every OPN member with packaged applications must seriously consider taking steps to become Exastack Ready, or Exastack Optimized at the first opportunity. That first step down the track is to talk to an expert on the OPN Portal, at the Oracle Partner Business Center or to discuss the next steps with the closest Oracle account manager. Oracle Exastack lab environments and other technical enablement resources are available for OPN members wishing to further their knowledge of Oracle Exastack and qualify their applications for Oracle Exastack Optimized. New Boot Camps and Guided Learning Paths (GLPs), tailored specifically for ISVs, are available for Oracle Exadata Database Machine, Oracle Exalogic Elastic Cloud, Oracle Linux, Oracle Solaris, Oracle Database, and Oracle WebLogic Server. More information about these GLPs and Boot Camps (including delivery dates and locations) are posted on the OPN Competency Center and corresponding OPN Knowledge Zones. Learn more about Oracle Exastack labs and ISV specific enablement resources. "Oracle Specialized partners are of course front-and-centre, with potential customers clearly directed to those partners and to Exadata Ready partners as a matter of priority." --More OpenWorld 2011 highlights for Oracle partners and customers Oracle Application Testing Suite 9.3 application testing solution for Web, SOA and Oracle Applications Oracle Application Express Release 4.1 improving the development of database-centric Web 2.0 applications and reports Oracle Unified Directory 11g helping customers manage the critical identity information that drives their business applications Oracle SOA Suite for healthcare integration Oracle Enterprise Pack for Eclipse 11g demonstrating continued commitment to the developer and open source communities Oracle Coherence 3.7.1, the latest release of the industry's leading distributed in-memory data grid Oracle Process Accelerators helping to simplify and accelerate time-to-value for customers' business process management initiatives Oracle's JD Edwards EnterpriseOne on the iPad meeting the increasingly mobile demands of today's workforces Oracle CRM On Demand Release 19 Innovation Pack introducing industry-leading hosted call centre and enterprise-marketing capabilities designed to drive further revenue and productivity while reducing costs and improving the customer experience Oracle's Primavera Portfolio Management 9 for businesses delivering on project portfolio goals with increased versatility, transparency and accuracy Oracle's PeopleSoft Human Capital Management (HCM) 9.1 On Demand Standard Edition helping customers manage their long-term investment in enterprise-wide business applications New versions of Oracle FLEXCUBE Universal Banking and Oracle FLEXCUBE Investor Servicing for Financial Institutions, as well as Oracle Financial Services Enterprise Case Management, Oracle Financial Services Pricing Management, Oracle Financial Management Analytics and Oracle Tax Analytics Oracle Utilities Network Management System 1.11 offering new modelling and analysis features to improve distribution-grid management for electric utilities Oracle Communications Network Charging and Control 4.4 helping communications service providers (CSPs) offer their customers more flexible charging options Plus many, many more technology announcements, enhancements, momentum news and community updates -- Oracle OpenWorld 2012 A date has already been set for Oracle OpenWorld 2012. Held once again in San Francisco, exhibitors, partners, customers and Oracle people will gather from 30 September until 4 November to meet, network and learn together with the rest of the global Oracle community. Register now for Oracle OpenWorld 2012 and save $$$! We'll reward your early planning for Oracle OpenWorld 2012 with reduced rates. Super Saver deals are now available! -- Back to the welcome page

    Read the article

  • Hello it’s your server calling

    - by GrumpyOldDBA
    This is nothing exciting but I've always found this startup procedure  very useful. All this simple procedure does is send you an email if the SQL Service Starts. If your Server is a cluster it will tell you which node you're on. -- On it's own this procedure can't actually be used as I route the output through another procedure, dbasp_SendMessage, this procedure routes a passed message to either a smtp email or a log table or both, the destination is set in a server config table...(read more)

    Read the article

  • Microsoft Windows HPC Server R2 Beta2

    Internally and unofficially we refer to this as "HPC Server v3" and its Beta2 became available last week. Read the full story on this blog post from Ryan and this one from Don. There has been a lot of excitement on the web for this release with coverage from last Wednesday here, here, here, here, here and here. Don't forget that Visual Studio 2010 makes it easy to develop for HPC Server including the MPI Cluster Debugger integration that I explained here and here. Comments about this...Did you know that DotNetSlackers also publishes .net articles written by top known .net Authors? We already have over 80 articles in several categories including Silverlight. Take a look: here.

    Read the article

  • Microsoft Windows HPC Server R2 Beta2

    - by Daniel Moth
    Internally and unofficially we refer to this as "HPC Server v3" and its Beta2 became available last week. Read the full story on this blog post from Ryan and this one from Don. There has been a lot of excitement on the web for this release with coverage from last Wednesday here, here, here, here, here and here. Don't forget that Visual Studio 2010 makes it easy to develop for HPC Server including the MPI Cluster Debugger integration that I explained here and here. Comments about this post welcome at the original blog.

    Read the article

  • Visual History for Chrome Maps Out Your Browser History in an Interactive Graph

    - by Jason Fitzpatrick
    Curious how your adventures on the web interweave? Visual History for Chrome maps out related web sites in your browsing history into an interactive chart–visualize your browsing over the last hours, days, or months. One of the interesting elements of Visual History is that it doesn’t simply link sites together via activated hyperlinks but by consecutive use within 20 minute increments–thus if you frequently hit up Gmail, Facebook, and Reddit first thing in the morning, they’ll all appear together in a usage cluster. Site can be organized by URL, sub-domain, or domain. Visual History is free, Chrome only. Visual History for Chrome [Chrome Web Store] HTG Explains: What The Windows Event Viewer Is and How You Can Use It HTG Explains: How Windows Uses The Task Scheduler for System Tasks HTG Explains: Why Do Hard Drives Show the Wrong Capacity in Windows?

    Read the article

  • Oracle Enterprise Data Quality: Ever Integration-ready

    - by Mala Narasimharajan
    It is closing in on a year now since Oracle’s acquisition of Datanomic, and the addition of Oracle Enterprise Data Quality (EDQ) to the Oracle software family. The big move has caused some big shifts in emphasis and some very encouraging excitement from the field.  To give an illustration, combined with a shameless promotion of how EDQ can help to give quick insights into your data, I did a quick Phrase Profile of the subject field of emails to the Global EDQ mailing list since it was set up last September. The results revealed a very clear theme:   Integration, Integration, Integration! As well as the important Siebel and Oracle Data Integrator (ODI) integrations, we have been asked about integration with a huge variety of Oracle applications, including EBS, Peoplesoft, CRM on Demand, Fusion, DRM, Endeca, RightNow, and more - and we have not stood still! While it would not have been possible to develop specific pre-integrations with all of the above within a year, we have developed a package of feature-rich out-of-the-box web services and batch processes that can be plugged into any application or middleware technology with ease. And with Siebel, they work out of the box. Oracle Enterprise Data Quality version 9.0.4 includes the Customer Data Services (CDS) pack – a ready set of standard processes with standard interfaces, to provide integrated: Address verification and cleansing  Individual matching Organization matching The services can are suitable for either Batch or Real-Time processing, and are enabled for international data, with simple configuration options driving the set of locale-specific dictionaries that are used. For example, large dictionaries are provided to support international name transcription and variant matching, including highly specialized handling for Arabic, Japanese, Chinese and Korean data. In total across all locales, CDS includes well over a million dictionary entries.   Excerpt from EDQ’s CDS Individual Name Standardization Dictionary CDS has been developed to replace the OEM of Informatica Identity Resolution (IIR) for attached Data Quality on the Oracle price list, but does this in a way that creates a ‘best of both worlds’ situation for customers, who can harness not only the out-of-the-box functionality of pre-packaged matching and standardization services, but also the flexibility of OEDQ if they want to customize the interfaces or the process logic, without having to learn more than one product. From a competitive point of view, we believe this stands us in good stead against our key competitors, including Informatica, who have separate ‘Identity Resolution’ and general DQ products, and IBM, who provide limited out-of-the-box capabilities (with a steep learning curve) in both their QualityStage data quality and Initiate matching products. Here is a brief guide to the main services provided in the pack: Address Verification and Standardization EDQ’s CDS Address Cleaning Process The Address Verification and Standardization service uses EDQ Address Verification (an OEM of Loqate software) to verify and clean addresses in either real-time or batch. The Address Verification processor is wrapped in an EDQ process – this adds significant capabilities over calling the underlying Address Verification API directly, specifically: Country-specific thresholds to determine when to accept the verification result (and therefore to change the input address) based on the confidence level of the API Optimization of address verification by pre-standardizing data where required Formatting of output addresses into the input address fields normally used by applications Adding descriptions of the address verification and geocoding return codes The process can then be used to provide real-time and batch address cleansing in any application; such as a simple web page calling address cleaning and geocoding as part of a check on individual data.     Duplicate Prevention Unlike Informatica Identity Resolution (IIR), EDQ uses stateless services for duplicate prevention to avoid issues caused by complex replication and synchronization of large volume customer data. When a record is added or updated in an application, the EDQ Cluster Key Generation service is called, and returns a number of key values. These are used to select other records (‘candidates’) that may match in the application data (which has been pre-seeded with keys using the same service). The ‘driving record’ (the new or updated record) is then presented along with all selected candidates to the EDQ Matching Service, which decides which of the candidates are a good match with the driving record, and scores them according to the strength of match. In this model, complex multi-locale EDQ techniques can be used to generate the keys and ensure that the right balance between performance and matching effectiveness is maintained, while ensuring that the application retains control of data integrity and transactional commits. The process is explained below: EDQ Duplicate Prevention Architecture Note that where the integration is with a hub, there may be an additional call to the Cluster Key Generation service if the master record has changed due to merges with other records (and therefore needs to have new key values generated before commit). Batch Matching In order to allow customers to use different match rules in batch to real-time, separate matching templates are provided for batch matching. For example, some customers want to minimize intervention in key user flows (such as adding new customers) in front end applications, but to conduct a more exhaustive match on a regular basis in the back office. The batch matching jobs are also used when migrating data between systems, and in this case normally a more precise (and automated) type of matching is required, in order to minimize the review work performed by Data Stewards.  In batch matching, data is captured into EDQ using its standard interfaces, and records are standardized, clustered and matched in an EDQ job before matches are written out. As with all EDQ jobs, batch matching may be called from Oracle Data Integrator (ODI) if required. When working with Siebel CRM (or master data in Siebel UCM), Siebel’s Data Quality Manager is used to instigate batch jobs, and a shared staging database is used to write records for matching and to consume match results. The CDS batch matching processes automatically adjust to Siebel’s ‘Full Match’ (match all records against each other) and ‘Incremental Match’ (match a subset of records against all of their selected candidates) modes. The Future The Customer Data Services Pack is an important part of the Oracle strategy for EDQ, offering a clear path to making Data Quality Assurance an integral part of enterprise applications, and providing a strong value proposition for adopting EDQ. We are planning various additions and improvements, including: An out-of-the-box Data Quality Dashboard Even more comprehensive international data handling Address search (suggesting multiple results) Integrated address matching The EDQ Customer Data Services Pack is part of the Enterprise Data Quality Media Pack, available for download at http://www.oracle.com/technetwork/middleware/oedq/downloads/index.html.

    Read the article

< Previous Page | 82 83 84 85 86 87 88 89 90 91 92 93  | Next Page >