Re: Using redo logs for a hot standby

JC (j.corbett_at_bbcnc.org.uk) wrote:
: Hi,
 

: I wonder if anyone could offer any advice on the following scenario.
 

: We have a t.p. system that uses an Oracle 7 database running under Open VMS
: on a DEC Alpha. This system MUST be available as near to 100% of the
: time as possible during working hours. The system is running in archivelog mode
: and volume shadowing is enabled at the OS level.
 

: However, the organisation wishes to protect a disaster situation ,eg a 747 landing on
: our computer room. A colleague has suggested a possible 'hot standby' situation:
: - having a second identical system at another site (which is already coonected via a WAN link)
: to which we copy the redo logs and apply them every 15 minutes. This sounds logical except the application of the logs on the second system
: every 15 minutes. Shoule we also copy over (backup) the control file from
: the primary system - would this cause the secondary system to realize that recovery is necessary?
: Has anyone come across such a solution before? (Snapshots are not appropriate for our setup)

the above system is definitely doable. I've set up one for a previous client and am in the process of designing another for my current client. the thing to keep in mind is that the the second db is forever in recovery mode, so you copy the control file (along with the datafiles) over to the secondary system once, start it up mounted and start doing a database manual recovery ('alter database recover using backup controlfile until cancel').

to automate the whole thing, write two daemon programs: one which resides on the primary system and copied archive logs over to the secondary system, and another daemon which resides on the secondary machine which would apply the newly received logs into the mounted db.

Johnny Chan
Indepedent Oracle Specialist Received on Fri Sep 15 1995 - 00:00:00 CEST