Re: Load Balancing VIP in SQL Server Clustering

Tech-Archive recommends: Fix windows errors by optimizing your registry

The problem is that you cannot guarantee that SQL Serer will be up 100% of
the time. If it goes down for even a second, all currently-running sessions
are disconnected and any transactions that were under way are rolled back.
Consequently, the app simply has to retry. Keep in mind that a network
interruption between the client and SQL Server can have the same effect.
Build the app to retry when it encounters a disconnect and you have made the
app itself fault tolerant.

--
Tom

----------------------------------------------------
Thomas A. Moreau, BSc, PhD, MCSE, MCDBA, MCITP, MCTS
SQL Server MVP
Toronto, ON Canada
https://mvp.support.microsoft.com/profile/Tom.Moreau


"Steve Kitley" <SteveKitley@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:D52C0143-57D9-4A97-BEBD-089A6C1128D1@xxxxxxxxxxxxxxxx
Hi Tom,

Thanks for your reply. So you do not think there is anyway to create fault
tolerance via a hardware load balancer with a virtual IP address?

Steve

"Tom Moreau" wrote:

There is a difference between fault tolerant and high availability. I
don't
know of any system that will persist open transactions in a high
availability system. Rather, the calling program should attempt to re-try
a
transaction if connectivity is lost.

--
Tom

----------------------------------------------------
Thomas A. Moreau, BSc, PhD, MCSE, MCDBA, MCITP, MCTS
SQL Server MVP
Toronto, ON Canada
https://mvp.support.microsoft.com/profile/Tom.Moreau


"Steve Kitley" <Steve Kitley@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:F1B8F3C7-2659-483B-A8B4-1BB9B3140D3B@xxxxxxxxxxxxxxxx
Hi,

From what I have read, SQL Server is per se not fault tolerant. In other
words, in a two node instance should one node fail the transactions in
progress will not automatically switchover and hit the node that is still
running, ie, there is a lag in the switching of the nodes. Does anybody
know that if we put a load balancer with a virtual ip address in front of
the
cluster would resolve this problem? Now, the load balancer can be either
a
software or hardware solution, but I believe a hardware load balancer
appliance would be better.





.

Relevant Pages

  • Re: Access 2007->SQL Server2005 "connection was forcibly closed",G
    ... I moved every table I was able to move to the SQL ... closed connections - but all of these errors are in the version which used ... the SQL Server 2000 and everything worked ... communication between ODBC (OLEDB and Native Client, ...
    (microsoft.public.sqlserver.connect)
  • Re: Unable to Apply SP4 to SQL 2000 Cluster (new Node)
    ... Rebuild the node in the failover cluster. ... Scenario 1" in SQL Server 2000 Books Online. ... This setup process updates to SP4 only the binaries on the new ...
    (microsoft.public.sqlserver.clustering)
  • Re: WSS 3.0 question
    ... I followed the advise given in removing WSS 3.0 etc, ... the server is complaining that the SQL service(?) was tempered with or corrupt. ... I may just instal the SQL server as I was going eventuall use it anyway. ... If WSUS 3.0 is installed, I would suggest you uninstall it and then you install WSS 3.0. ...
    (microsoft.public.windows.server.sbs)
  • Re: WSUS
    ... I'm not seeing performance issues with the full enchilada installed, and 25 users busy hitting SQL. ... WSUS isn't difficult to uninstall - if you have WSUS v2 (installed with SBS R2) uninstall R2 from add/remove programs. ... How can anyone work with 4 instances of SQL Server on the same box? ...
    (microsoft.public.windows.server.sbs)
  • Re: SQL Resets
    ... If it were SQL that was falling short, ... The default backlog for SQL Server is 5. ... System.InvalidOperationException: Internal connection fatal error. ... From time to time, under heavy loads, we are getting resets at ...
    (microsoft.public.sqlserver.connect)