2 Replies Latest reply on Apr 12, 2013 9:49 AM by bevans88

    Cross Site Replication - Valid Usage?

    bevans88 Newbie



      I'm currently looking at using cross-site replication and have noticed some problems with using it in the way we envisaged (which I understand may not be the intended use when it was implemented/designed). I have written a brief use case below which roughly covers our intended usage:


      Use Case:


      • Three sites; A, B, C, configured as backups of each other (B and C backup A etc.)
      • Data can be accessed on all sites, eg. site B has cached copies/access to Site A and Cs data.
      • If Site A is shutdown, when restarted, all of its remotely backed up data is restored to the cache(s).
        • The cache(s) being backed up will all have the same name and so all sites will store backed up data in the same cache as the sites 'local' data.




      • If Site A has local data in the cache (which is successfully backed up), upon shutting down and restarting Site A it does not retrieve this data back.
      • If a site adds data to its local cache while the backup sites are down/inaccessible, this data is not backed up to these sites when they become available again.


      Is it possible to solve this through the use of cross-site replication?


      tldr; is there a way to force a site to get the data from it's back-up caches and reload it into its local cache?


      I am also looking at cluster re-merging and how Infnispan handles this (I can't seem to find anything in the documentation.):


      1. Site A adds data to its cache, which is backed up onto Site B
      2. Comms between the sites is lost.
      3. Site A updates the data which is backed up on Site B.
      4. Comms are restored


      In this instance which version of the data would be in the caches? Also, what would happen in the reverse scenario (Site B updates the data)?