none
failover to backup web service RRS feed

  • Question

  • We have a failover site to host our web service for a company mobile app if our main site goes down. To implement support for it, I coded something like this:

    Try
      ws.Login(username, pwd)
    Catch
      ws.Url = Config.GetSetting("backupUrl")
      ws.Login(username, pwd)
    End Try

    See any hazards? Thanks.

    Sunday, January 28, 2007 12:07 AM

Answers

  • Why have a 'backup' service when you could load balance 2 or more servers and therefore not have to worry about fail over servers from a code perspective.

    HTH

    Ollie Riches

    Sunday, January 28, 2007 7:41 PM
  • You should probably be more selective in the catch logic.  You want to try again if there's a communications failure but probably not if the login failed for bad credentials or other errors.
    Monday, January 29, 2007 1:58 AM
  • DRP solution is usually a multi-tiered effort - moving to a remote site is usually only done when you have a catastrophy - because of issues like data synchronization and freshness -Even though it depends on your particular business, you would probably find that you cannot synchronize the whole data between the two sites at the same level of synchronization

    You may still want to add NLB or other clustering solution on your "main" site and handle the rerouting of services outside your code in cases of a site fail-over

    Arnon

    Tuesday, February 13, 2007 10:58 PM

All replies

  • Why have a 'backup' service when you could load balance 2 or more servers and therefore not have to worry about fail over servers from a code perspective.

    HTH

    Ollie Riches

    Sunday, January 28, 2007 7:41 PM
  • You should probably be more selective in the catch logic.  You want to try again if there's a communications failure but probably not if the login failed for bad credentials or other errors.
    Monday, January 29, 2007 1:58 AM
  • Thanks for the comments. (Perhaps I should have mentioned, the servers are in separate cities, for disaster hardening).
    Tuesday, February 6, 2007 7:21 PM
  • Thanks for the feedback!
    Tuesday, February 13, 2007 8:47 PM
  • DRP solution is usually a multi-tiered effort - moving to a remote site is usually only done when you have a catastrophy - because of issues like data synchronization and freshness -Even though it depends on your particular business, you would probably find that you cannot synchronize the whole data between the two sites at the same level of synchronization

    You may still want to add NLB or other clustering solution on your "main" site and handle the rerouting of services outside your code in cases of a site fail-over

    Arnon

    Tuesday, February 13, 2007 10:58 PM
  • Do the server share a SAN across data centres?  If so you should be able to replicate over that; where is your DB?
    Thursday, February 15, 2007 2:09 AM