Post reply

Choosing replication type and preparing for it

  • September 29, 2013 at 3:53 am

    #283158

    Hi all,

    We have 15 servers (in different regions) and each need to work with subset of data and exchange modifications with central the server.

    I have chosen merge replication with dynamic filter to partition data, therefore central server acts as publisher and 15 servers as subscriber.

    The problem is that data is mainly generated at subscribers and when I want to initialize snapshot, the central server has nothing to send to subscriber. Besides, if I sync them manually for startup, it will be difficult if in the future we require to reinitialize particular subscriber because the changes at the subscriber happens a lot.

    Another idea is to use central subscriber model but I'm not sure if this is the right choice.

    Please advise me to choose right approach.

    Many thanks in advance,

    Leila

  • September 30, 2013 at 6:27 am

    #1654512

    What I understood from above post is "Data gnerated at subscriber from 15 sites and replicated to central server". If this is the case central subscriber is the ideal option. And yes, you can setup 15 sites as Publishers. However if the data also publish from Central server than you can set up merge replication.

    Still provide more information about the data flow and what exact issue you are having so that people can help you.

    HTH

    ---------------------------------------------------
    "Thare are only 10 types of people in the world:
    Those who understand binary, and those who don't."

  • October 1, 2013 at 1:47 am

    #1654793

    Thanks for your reply! 🙂

    It is exactly what you understood! By using merge replication, I think we need to configure 15 servers as subscriber. This way there will be problem for startup snapshots and later re-initialization if required.

    For startup snapshot, central server is empty and has nothing to send as snapshot to subscribers. Each subscriber must send the initial data to publisher(how!? SSIS? Linked server?).

    If in the future we need to re-initialize particular subscriber, all data related to that subscriber(partition) must be deleted at publisher, and send data from subscriber and re-establish replication.

    What can be good solution?

    Thanks again!

Viewing 3 posts - 1 through 2 (of 2 total)

You must be logged in to reply to this topic. Login to reply