Re: Preparing for HA solution

[Jan Megannon <jmegannon@xxxxxxxxxxxxx>]

Hi Oliver,

See inline.

ouuch@t-online.de wrote:
> We'd like to have a hot-standby and be able to switch over in less than 1 hour.
> Currently, it looks like DataMirror will be the HA-provider.
There is a product that we use, but this info I will sent directly to 
you. Don't want to appear to be marketing in this forum ;-)
With this product, you can switch over to your HA machine very quickly. 
Dependng on the functionality that you purchase, you could then 
replicate all the changes from the HA system back to the Prod system 
once that is back up and running. This is assuming that your Prod 
machine wasn't trashed in the disaster and depends on the degree of the 
disaster.

> AFAIK, normal solutions have the stand-by machine sitting mostly idle in  its own,
> separate machine room with separate network connection and all that stuff.
> So, could the stand-by machine be used for any real work (maybe LPAR as a test or
> development machine)? 
Depending on the HA product you install (and the way you install it), 
your replication may be anything from almost instantaneous to a bit 
delayed. You could use your HA system for running your queries, reports, 
etc. You could also use it in the following manner:
	Normal production day (remote journaling and replication active)
	End remote journaling (replication)
	Run backups of production data on the HA system
	Run Day End on Prod system at the same time
	Restart remote journaling (replication) on completion of backups

This allows your prod system to remain available 24/7. This type of 
implementation depends on how much you trust the implementation and the 
HA product!

You could also do the following at monthend:
	End replication
	Backup production data on the HA system
	Start replication
	Run Month End
	Check that everything is replicated
	End replication (when complete)
	Continue production on Prod system
	Run Month End reports on HA system
	Restart replication once reports are complete
There are certain things to be aware of here, but these should all come 
out in the design and planning stages of your HA implementation.

> Are there any reports (in-depth) on how this was done. What are the upcoming problems?
> What should be done in preparation (or can be done)?
>
> Thanks for sharing your experiences,
>
> Oliver
Just some thoughts that work.

Jan.