There are a number of options:
- Create a db link and only grant select to the tables. This will impact
prod performance and probably get the prod dba upset as well as violate your
internal corporate policies. Not much good if you want to perform updates
from the test application. The advantage is that, under controlled
conditions, no data needs to be copied.
- exp/imp the dynamic tables and create select via a db link on the
metadata. Less impact to prod, less data to move. May keep the prod dba
happy - maybe not.
- Cold backup/restore. Very fast but requires two hosts with similar
filsystem layouts (everyone has implemented OFA haven't they?) as you can't
have two instances with the same db SID on the one host. Copy the files over
your FDDI backup ring any time of day without impact to prod network.
Disadvantage is that your have to take prod down to do the backup.
- Hot backup/restore. As above but don't take the prod db down. This works
a treat. Trivial to implement and manage.
- Just get an export of the prod schema and load it into your existing db
instance on your test box. i.e. your test schema is "test" and your prod
schema is "prod". Maybe you will have to use "fromuser" and "touser"
parameters of import. This is most likely the one that will keep everyone
happy. Problems are - performance impact on prod while you are taking the
export, data consistency (difficult in a prod environment without blowing
rollbacks - snapshot too old will be your error during export). Export to a
pipe via a remote shell (remsh) so that you don't have to stage intermediate
files. In fact, if you know your way around UNIX you can exp to a pipe to a
remote shell to stdout to a pipe to imp to get the data directly from one db
to another.
- Buy EMC disk frames and use EMC's SRDF to very quickly copy (track by
track) in hot backup mode. Not for the feint hearted or the budget
challenged but it can move hundreds of Gigabytes in less than an hour. Very
neat. I've experience of this for over a year and the main problems were
keeping Solaris patches and EMC microcode in synch.
There are probably a few more options, but hey, that's why you posted to a
news group. I'll leave room for someone else.
...neil {actually: neil [dot] hulin [at] litech [dot] freeserve [dot] co
[dot] uk}
Received on Fri Nov 13 1998 - 13:12:20 CST
