|
|
 |
 |
 |
 |
backing-up with rman and expdump
Why would one need to backup entire database: sysaux, system, etc with rman, if all I got is tables, functions, indexes, triggers and sequences? I do not see any difference between restoring db from full rman backup, and from expdmp.
Joe <joe1 @tlen.pl> wrote in news:1178766760.251353.119110 @h2g2000hsg.googlegroups.com: > Why would one need to backup entire database: sysaux, system, etc with > rman, if all I got is tables, functions, indexes, triggers and > sequences? > I do not see any difference between restoring db from full rman > backup, and from expdmp.
Have you ever actually did a full Production restoration via impdmp? I suspect NOT. -----------------------------------------------Reply-----------------------------------------------
On 9 May 2007 20:12:40 -0700, Joe <joe1 @tlen.pl> wrote: >Why would one need to backup entire database: sysaux, system, etc with >rman, if all I got is tables, functions, indexes, triggers and >sequences? >I do not see any difference between restoring db from full rman >backup, and from expdmp.
expdmp is a logical backup. Nothing more than a series of insert statements. So importing doesn't guarantee all records return to their original locations. Consequently you won't be able to recover the database (= apply (archived) redologs) after impdmp. You don't make an expdmp every 2 minutes, do you? -- Sybrand Bakker Senior Oracle DBA
-----------------------------------------------Reply-----------------------------------------------
On May 10, 12:59 am, sybra @hccnet.nl wrote: > expdmp is a logical backup. Nothing more than a series of insert > statements. So importing doesn't guarantee all records return to their > original locations. Consequently you won't be able to recover the > database (= apply (archived) redologs) after impdmp. > You don't make an expdmp every 2 minutes, do you?
I do expdmp 2 times a day, noon and evening, after 6pm. I know I wont be able to use redologs. Why would my records not get to the right place, just like they were before? i do not have stored procedure in db. -----------------------------------------------Reply-----------------------------------------------
On May 10, 10:03 am, Joe <joe1 @tlen.pl> wrote: > On May 10, 12:59 am, sybra @hccnet.nl wrote: > > expdmp is a logical backup. Nothing more than a series of insert > > statements. So importing doesn't guarantee all records return to their > > original locations. Consequently you won't be able to recover the > > database (= apply (archived) redologs) after impdmp. > > You don't make an expdmp every 2 minutes, do you? > I do expdmp 2 times a day, noon and evening, after 6pm. > I know I wont be able to use redologs. > Why would my records not get to the right place, just like they were > before? > i do not have stored procedure in db.
By taking to exports you have two images at certain points in time, but nothing in between or afterwards. So, if you exported at 6pm and your database crashes at 10pm you will only be able to recover an image as of 6pm (all changes to the database after 6pm are lost for good). In short: Exp/imp (data pump) = snapshots/images at a specific times RMAN + archive logs = snapshots + everything that "happened" in between (changes on data, objects, etc)
-----------------------------------------------Reply-----------------------------------------------
On May 10, 10:14 am, Valentin Minzatu <valentinminz @yahoo.com> wrote: > In short: > Exp/imp (data pump) = snapshots/images at a specific times > RMAN + archive logs = snapshots + everything that "happened" in > between (changes on data, objects, etc)
Valentin, if I make backup through RMAN, it will also be up to the certain point (exact to the moment that I've execute RMAN)
-----------------------------------------------Reply-----------------------------------------------
On Thu, 10 May 2007 07:29:05 -0700, Joe wrote: > Valentin, > if I make backup through RMAN, it will also be up to the certain point > (exact to the moment that I've execute RMAN)
With RMAN, you can roll forward. I would suggest reading the Rob Freeman's RMAN book. You might find it enlightening. -- http://www.mladen-gogala.com
-----------------------------------------------Reply-----------------------------------------------
On May 10, 10:29 am, Joe <joe1 @tlen.pl> wrote: > On May 10, 10:14 am, Valentin Minzatu <valentinminz @yahoo.com> > wrote: > > In short: > > Exp/imp (data pump) = snapshots/images at a specific times > > RMAN + archive logs = snapshots + everything that "happened" in > > between (changes on data, objects, etc) > Valentin, > if I make backup through RMAN, it will also be up to the certain point > (exact to the moment that I've execute RMAN)
Please read above: RMAN ****+ archive logs **** = snapshots + everything that "happened" in
> > between (changes on data, objects, etc)
On May 9, 11:12 pm, Joe <joe1 @tlen.pl> wrote: > Why would one need to backup entire database: sysaux, system, etc with > rman, if all I got is tables, functions, indexes, triggers and > sequences? > I do not see any difference between restoring db from full rman > backup, and from expdmp.
The difference is you are not "restoring a database " with expdmp, you are recreating objects and running inserts. expdmp is not "restoring the database" it is recreating objects within a "new" database. RMAN is a true file level "Database" restore. It also checks for block corruption during backup. It can be optimized on multiple levels, works directly with tape storage and keeps track of backups whether in catalog or in controlfile. It has no problem with XLM types or lobs and can backup and restore to an audit based point-in-time. Try backing up an xmldb with expdmp. EscVector Databases are forever, instances are ephemeral.... :)
-----------------------------------------------Reply-----------------------------------------------
Joe wrote: > On May 10, 10:14 am, Valentin Minzatu <valentinminz @yahoo.com> > wrote: >> In short: >> Exp/imp (data pump) = snapshots/images at a specific times >> RMAN + archive logs = snapshots + everything that "happened" in >> between (changes on data, objects, etc) > Valentin, > if I make backup through RMAN, it will also be up to the certain point > (exact to the moment that I've execute RMAN)
Your assumptions are not supported by the facts. What you have now is not a backup and can not be used as a mechanism for restoration and recovery. It is time to read the RMAN concepts docs. From where I sit often advising companies on hiring and policies. Anyone not using RMAN in 9i+ is unemployable. -- Daniel A. Morgan University of Washington damor@x.washington.edu (replace x with u to respond) Puget Sound Oracle Users Group www.psoug.org
-----------------------------------------------Reply-----------------------------------------------
On May 10, 7:03 am, Joe <joe1 @tlen.pl> wrote: > On May 10, 12:59 am, sybra @hccnet.nl wrote: > > expdmp is a logical backup. Nothing more than a series of insert > > statements. So importing doesn't guarantee all records return to their > > original locations. Consequently you won't be able to recover the > > database (= apply (archived) redologs) after impdmp. > > You don't make an expdmp every 2 minutes, do you? > I do expdmp 2 times a day, noon and evening, after 6pm. > I know I wont be able to use redologs. > Why would my records not get to the right place, just like they were > before? > i do not have stored procedure in db.
You really need to read the backup and recovery basics manual. There you find things like this, right at the beginning: "Physical backups are the foundation of any sound backup and recovery strategy. Logical backups are a useful supplement to physical backups in many circumstances but are not sufficient protection against data loss without physical backups." Now, what they might mean here is not losing any transactional data, and for your organization it may seem sufficient if they "only" lose half a days work. (From what I've seen, such a judgment usually comes from faulty risk analysis.) But... What is also implicit is that you need redundancy in any recovery plan. You need to remember, any given backup, logical or physical, may be unusable for a variety of reasons. With a logical backup, if one is bad, you can only go to the previous. If that one is bad, the one before that. And so on until you realize you are TSOL. With physical backups, it is possible to have multiple copies of archived logs in various backups (as well as online), so it may be possible to go to an earlier backup and apply logs as much as desired. In addition, as others have alluded to, RMAN has additional capabilities for corruption detection and repair. Beyond that, there are newer features based on archive logging that allow recovery without even bothering with the backups. There is also the occasional glitch doing an import. The logical backups are really for moving data around (that would explain the data pump name, right?), and in some cases recover some data from a previous time. They are not designed to be depended on as a database backup. We can't be any clearer than that: read the manuals. That said, I use exp/imp all the time for various things, it is very useful to have recent exports as a matter of course. It's just not for primary backup purposes. You might also be taking note of any effects on your undo if you are exporting during transactional time. jg -- @home.com is bogus. "When a meteorologist says 'we'll be back with tomorrow's forecast, right after this,' he's trying to sneak away to the bathroom. Some of the more relaxed ones are actually catheterized, with a tube running down their pant leg and into a bottle strapped just above the ankle. The camera never shows a meteorologist below the knee, and now you know why." - Dr. Science
-----------------------------------------------Reply-----------------------------------------------
On Thu, 10 May 2007 09:14:59 -0700, DA Morgan wrote: > Your assumptions are not supported by the facts.
Is this the reason to discard a beautiful assumption? Not so long ago, Stephen Colbert had a very amusing diatribe against "factists". If we introduce the requirement for the facts and proofs, many things will have to be discarded. We should leave things as they are now, intelligently designed. -- http://www.mladen-gogala.com
|
 |
 |
 |
 |
|