Hello guys,   I would like to get information about Non ODBC Data Connections between Dataflux 2.6 and Oracle. Are there any kind of way to use the SAS Access to Oracle in a Dataflux code? Or Dataflux can only access Oracle database by ODBC or Fed Server? Thanks in advance for any info. ... View more

Hello guys,   I would like to get information on the minimum recommended amount of cores and memory for the SAS Federation Server installation.   I tried to look for this information in the product Sytem Requirements pages but I did not find these references. http://support.sas.com/documentation/installcenter/en/ikdffedsvrsr/69017/HTML/default/index.html We are using an Linux for x64 platform for a SAS Federation Server 4.2  Thanks in advance for any info. ... View more

Hello guys! I would like to try to get guidance on the QKB Management process, regarding Customizations and Updating Versions in the following scenario I have a QKB Portuguese BR (eg CI24) that has undergone customizations (insertion of new definitions to meet specific needs). Once the SAS version is available to CI26, we need to define a procedure standard, recommended by SAS, to perform an update of my QKB to version 26, taking into account: - Without losing the customizations implemented - Ensuring the compatibility of code already implemented, and in production, which uses the original structure of QKB ​​CI 24 In this way I would like to try to get some answers to the questions I am having regarding: A) Please clarify how to use the QKB Merge Tool   - Which QKB should we use as Source, the new QKB CI26 or the old QKB CI24 Customized?   - Do the created files (backup 001) refer to which "source"? Source or Destination QKB?   - Is there a difference in the merge result if we reverse the order of the QKBs in the process? What are they?   - Would anyone have any examples of using the command line method for the Merge Tool (cmtgui.exe), by doing a full merge between 2 QKBs and also a partial merge (just a few files)?   - Is there any specific recommendation (points of attention) for use of the Merge Tool command line, when we perform a partial merge?   - In order to ensure the compatibility of previously implemented code (Jobs, Process Jobs of Dataflux, etc ...), is the Merge Tool the correct option for updating a previously customized version of QKB?   B) How to use the QKB Difference Viewer tool to support the upgrade of a previously customized version of QKB?   - Would anyone have an example of a comparison of a Custom QKB vs. Original QKB of the same or higher version?   C) What are the impacts of instead of making a merge, simply make an import of the custom definitions in QKB CI24 for the new QKB CI26?   - Will there be a need to recode Jobs and Process Jobs, etc. from Dataflux, due to possible changes in the structure of the definitions implemented by SAS in its new versions of QKB?   - Is there any tool that will refer us to which Jobs and Process Jobs and other previously implemented elements need to be recoded (adjusted)? I know there are many questions and I thank you in advance for any collaboration or clarification on the subject. Regards, Luís Leite ... View more

Hi Guys,   We´re using Dataflux Data Management 2.6 to make a join sequence (Left Join) for 8 Fixed Width files (see the JOB Image attached):   File 1 (VINCULOS) 72 Million Lines (232 bytes each) - The Base File File 2 (REMUNE) 600 Million Lines (48 bytes each) File 3 (DECIMO) 100 Million Lines (42 bytes each) File 4 (HORAS) 190 Million Lines (34 bytes each) File 5 (CONTRI) 80 Million Lines (60 bytes each) File 6 (AFASTA) 8 Million Lines (88 bytes each) File 7 (DESLIGA) 72 Million Lines (116 bytes each) File 8 (TRABAL) 72 Million Lines (164 bytes each)   Files from 2 to 5 are submitted to a sort and a Surviving Record Identification, that will insure a maximum of 72 million lines for each one, before the Join.   As you can see in the images attached: - The Expresion red nodes are log counters for each 1 million lines read, - The orange nodes are for Sorting, - The Expression green nodes creates a few attributes, - The Surviving grey nodes uses Filed Rules based on "not Null" for some attributes, with no previous output, - The Blue Data Joining nodes are one based attribute Left Joins, with no memory load option, - The Output Fixed File Purple Node has 72 million lines, with 785 bytes each   Using that design, the job runs in 5.5 days in a Dataflux Data Management Server with 128 GB RAM, 20 cores and 13 TB free of disk.   We´re using, as primary definition, the server default values for JOINBYTES, CHUNKSIZE, BASE/SORTBYTES, BASE/SORTTEMP and BASE/SORTTHREADS.   So, I´d like to ask:   - Is that successive join design the right approach to this situation? Separating the Joins and creating intermediate files would be a better performance solution? - Based on the server configuration, what values for JOINBYTES, CHUNKSIZE, BASE/SORTBYTES, BASE/SORTTEMP and BASE/SORTTHREADS should I may consider? - Any suggestions to improve performance for this Job?   Thank you in advance for any contribution to improving the performance of this JOB.   Luis Leite ... View more