How do you implement 'resume' feature for file sync?

Locked How do you implement 'resume' feature for file sync?

  • 15. září 2009 7:07
     
     
    Hi All,

    I'm brand new to Sync, feature that I will need is File Sync in MSF.

    1. I'm synching very big files (video) in every sync, is it possible to support 'resume' when connection is lost in the middle of transferring?

    2. My sync client is about 5000 servers, what suggestion do you have for me to do file sync between these clients, do I need to enumerate each client and sync with them one by one, or FileSync in anyway have support for multicasting file sync?

    Thank you very much for the answer.
    • Přesunutý Max Wang_Chinasoft 20. dubna 2011 21:46 Forum consolidation (From:SyncFx - Technical Discussion [ReadOnly])
    •  

Všechny reakce

  • 17. září 2009 20:15
     
     
    Hi, hanifku.
    Per question 1 above, I don't think there's a way of resuming the transfer of the file itself.
    If the file is not transferred that means the change is not applied, and the metadata is not updated, so in next sync it will be re-detected, and reapplied.


    Adrian
  • 18. září 2009 21:50
    Přispěvatel
     
     Odpovědět
    Hi -

    1. There is no way to resume sync'ing of a single file when using the FileSync in MSF. Files that were completely sync'ed the previous time will not be sync'ed the next time but anything in the middle of a sync will be restarted. We have no extensibility points that will let you add support for this at this time.

    2. There is no inherent support for multicasting file sync, so these clients would need to sync with each other.

    What is your scenario that you have ~ 5000 clients? We have not seen too many scenarios which need that level of peers.

    Thanks
    Deepa
    Deepa ( Microsoft Sync Framework)
  • 14. prosince 2009 4:00
     
     
    Multicasting support will be good one.  I do have the same scenario say 2-5 different servers for load balancing (as a cost effective solution copying files to different machines).  Currently we perform just by running sych from each other.  Do you have any advice to implement multicasting by ourself instead of synching each other?

    Thanks.