Content Management and Capture

 View Only
  • 1.  DataCap 9.1.9 System requirements

    IBM Champion
    Posted Wed September 21, 2022 06:13 AM
    Hi,

    We are currently planning to deploy a standard Datacap environnement (3 servers as shown in the redbook https://www.redbooks.ibm.com/redbooks/pdfs/sg247969.pdf page 175), as per the system requirements I couldn't find details about CPU sizing. The only system requirement I've found is the SPCR : IBM Software Product Compatibility Reports

    Unfortunately there is only memory and disk requirement, can someone share a sample minimum sizing for Datacap for a typical environment. If possible what are the rule of thumbs to size the resources.

    P.S : We are planning a deployment on VMware 

    Thank you

    ------------------------------
    Largou walid
    ------------------------------


  • 2.  RE: DataCap 9.1.9 System requirements

    Posted Thu September 22, 2022 02:37 PM

    Hello Largou,

    The link in the compatibility report are bare minimum.  It all depend on the volume and the type of document you are processing.
    The key here is ensure that production required datacap to be in a distributed system.  Follow the redbook for vertical or horizontal scaling.
    In 9.19 they introduce 64 bit action library which require a little more memory IMO.

    Since it's a VMware environment you can expand as needed. 

    Rulerunner rule of thumbs is one core per thread.  You can add more thread to core provided cpu does not exceed 80%.
    I also recommend keeping your DB small since it's very chatty with the DB.
    https://www.ibm.com/support/pages/how-clean-out-large-volumes-old-and-stale-records-datacap-engine-database-tables

    Hope this helps.



    ------------------------------
    Blue Devil
    ------------------------------



  • 3.  RE: DataCap 9.1.9 System requirements

    Posted Thu September 29, 2022 10:17 AM
    Core to Thread ratio is 1:1.5 according to InfoCenter.

    "The maximum number of threads is 150% of the number of processors."
    https://www.ibm.com/docs/en/datacap/9.1.9?topic=applications-rulerunner-thread-configuration

    I think this ratio can be 1:1 depending on how busy a Ruleruner thread becomes.
    It depends on scan input amount and also how loaded your application is with actions.

    Memory is, a number of thread X 4GB, in my thinking. This is because a ruleruner thread is 32bit process so it can use up to 4GB each.
    I will give 6 core ruleruner server 32 GB. I think the disk size can be something like 200GB.
    Unless you want keep all logs, periodically deleting them saves disk space.
    In my project batch folder is stored in outside disk. It is something like 2TB size.

    Hope this helps.


    ------------------------------
    dsakai
    ------------------------------



  • 4.  RE: DataCap 9.1.9 System requirements

    Posted Tue December 20, 2022 01:14 PM
    FYI:  The 1::1.5 CPU "rule-of-thumb" was originally created for real servers, not VM's.   For VM's you need to know how many real CPU are on the Hypervisor (HV) (VMWare or Hyper-V) and how many LCPU each client VM has.  Add all of the client LCPU up and divide that into the number of real CPU's on the HV. If lower than .30 you are diluting the CPU power to the client VM's too much and you are going to have performance issues with DC RR servers in those RR VM's.   
    So for VM's I'd recommend sticking to a 1:1 LCPU ratio for the RR servers for higher volumes.

    IBM does have a DC sizing spreadsheet for calculating MIN CPU/MEM/Storage.  It is in the IBM Internal Capture Center of Competency box area.  You can ask your DC sales people to grab and use that to help you size your system.  

    At a high level for a PROD DC system you would need the following:

    BIG NOTE:  If a DC app routinely handles very large documents > 1 GB or 1,000's of pages
     and or if batches typically take > 4 hours to process,  then using a load balancer at the TMS level may not be right for your site. In this case create a PRIMARY and SECONDARY TMS server listing for each TMS server and try to split up the heaviest DC apps between the two TMS servers as the primary.  Simple, but effective.

    MIN PROD Config:
    TMS1
    TMS2

    FARMED:
    DC WEB1  wTM/TMWeb
    DC WEB2  wTM

    WAS ND 2+ nodes
     DC Navigator

    Additions:
    DC WEB 3  FP (Add a separate Finger Print (FP) server if you use over 75ish FP's.  If you have hundreds consider adding more FP servers for perf.)

    A
    WIN UNC Share for all DC apps and batch locations - can use multiple shares for very high volumes...  .A real SQL DB:  Db2, Oracle, SQL SVR with only ONE schema per DC app.  (Trivia: The docs were always written for TEST DC environments and the 3 DB's/schemas per DC app was ONLY meant for the little Access/Jet DB's.  Not for real ones that can handle more than one thread to the DB... So do your DBA's a favor and only use one schema per DC app plz - this actually performs slightly faster also ;-)

    ------------------------------
    Doug Short
    ------------------------------



  • 5.  RE: DataCap 9.1.9 System requirements

    Posted Fri April 28, 2023 10:23 AM

    Forgot the 2 needed Rulerunner (RR) servers.  Also I got the main DC 9.1.9 page to get updated to much better reflect the "one DC app, one DB schema" recommendation.  https://www.ibm.com/docs/en/datacap/9.1.9?topic=databases-configuring-application-use-database



    ------------------------------
    Doug Short
    ------------------------------