Re: Doubts on Data Guard and Replication

BlackBeltDBA wrote:
[snip]

> Switching to a standby database does incur *some* downtime. It's not an
> automatic event. The application will probably fail and need to be
> restarted once the standby comes back online. If you can not afford any
> downtime you will probably want RAC.

Could we perhaps kill this myth before it gets out of hand? There *is* downtime with RAC. Blocks that require re-mastering, as well as blocks which require instance recovery, will not be available until the remastering and the recovery have taken place. Your select will hang/pause/stall/whatever other word you'd like to use for an appreciable and measurable time until those things have happened. That means downtime in the sense of "Where's My Database Gone?!?"

It may only be seconds. It may be some tens of seconds. But it exists.

> Using Oracle's Transparent
> Application Failover (TAF), applications can fail over to another
> instance transparent to the end user.

It is important to qualify that statement. If you want it truly transparent (ie, where the user has to do absolutely nothing and if they'd been chatting with a colleague about last night's Survivor they wouldn't have noticed anything), then that requires the use of the SELECT method in the tnsnames (or its equivalent). Which kind of gives the game away that it only works for selects, and not DML (though I believe that has either changed in 10g or is about to... I can't recall which). OCI also has to be in the picture somewhere. And it only works when it does because it places a load on the client as results are returned.

> Alternatively, you can failover
> to another peer in your replicated environment. However, there may be
> some data loss that has to be handled.

There needn't be in Data Guard, but I think you covered that with the 'synchronous replication' stuff you mentioned earlier and which I snipped.

> If a few minutes of downtime is
> acceptable, a server cluster might be sufficient. If you can afford a
> longer outage then a standby database should be sufficient.

Don't disagree with that summary at all. I just don't like this "RAC is zero downtime" myth that is brewing in all quarters. It just ain't so.

Regards
HJR Received on Mon Nov 22 2004 - 21:06:52 CST