Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
Ive left my issues to the Management and DBA's - They will look at it "when time permits". NetBackup and myself have moved away from this !! Left it down to them to "manage" it so I will just grab a flatfile backup. Far as I am concerned I covered my back ! -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Lightner, Jeff Sent: 29 September 2011 18:59 To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) It might not be your fault if they did it anyway but that might not keep them from blaming you. Apparently your career has been such that you've never had to work with PHBs or organizational recalcitrance. I daresay most of us haven't been as lucky as you. Also implying you can quickly find another job also ignores realities that sometimes occur (e.g. the tech bust of 2002 or the economic meltdown of 2008/2009). Sometimes you just have to grin and bear it. -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Martin, Jonathan Sent: Thursday, September 29, 2011 1:18 PM To: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) Wow... sounds like accountability is in short supply. I work in I.T., it is not my job to tell management how to do theirs. If I was an electrician, and I told management they shouldn't plug more than 30A into a 30A outlet, and they did it anyway and then the breaker tripped, it wouldn't be my fault. And if they somehow caught the building on fire, I would still sleep well at night. I like to think I work together with management to analyze complex scenarios, explain them in simple terms, recommend paths for improvement, and let management decide what to do. Granted, every work environment is different. But if I had to save copies of CYA emails so I didn't lose my job over poor management decisions, that would be my "resume generating event." -Jonathan -Original Message- From: Lightner, Jeff [mailto:jlight...@water.com] Sent: Thursday, September 29, 2011 11:47 AM To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Question on DB online backups (Wayne T Smith) Often enough "marginally important" becomes "critically important" only after it fails. Usually that is when the admin faces the firing squad because management is never going to admit THEY didn't see it as important before the failure. If it is important enough to backup then it is important enough to test the restore. Telling folks later that it didn't seem important to you is NOT going to be the correct answer especially if they instructed you to do the backup in the first place. The best you can do is to write an email to your management explaining the issues and keep a copy of it (and better yet any response) for CYA. Even then it doesn't always help. In one environment we repeatedly told management about lack of backup being an issue and they repeatedly let us know that they didn't see it as an issue. When it finally failed the response we got back from on high was "You weren't forceful enough in saying it was important." -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Martin, Jonathan Sent: Thursday, September 29, 2011 10:33 AM To: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) While I agree that someone could face the firing squad if a certain database backed critical application failed and there was no restore capability, this may not be *that* application. Let's face it, if this application isn't critical enough to bring in a DBA, even as a contractor, and they didn't pay for implementation services with backups, then this application may just be marginally important. I would come up with and test some sort of restore scenario and set your management's expectations. That scenario does not need to be 24x7 snapshot based point in time restore with tape backup to provide recovery. Perhaps you take down the DB once a week to do a full, and the expectation is that if the database goes down you can lose up to a weeks' worth of data. Perhaps you supplement with snapshots, or database exports. Either way, Wayne's write-up is a great place to start. -Jonathan -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of David McMullin Sent: Thursday, September 29, 2011 9:57 AM To: veritas-bu@mailman.eng.auburn.edu Cc: wtsm...@maine.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wa
Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
It might not be your fault if they did it anyway but that might not keep them from blaming you. Apparently your career has been such that you've never had to work with PHBs or organizational recalcitrance. I daresay most of us haven't been as lucky as you. Also implying you can quickly find another job also ignores realities that sometimes occur (e.g. the tech bust of 2002 or the economic meltdown of 2008/2009). Sometimes you just have to grin and bear it. -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Martin, Jonathan Sent: Thursday, September 29, 2011 1:18 PM To: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) Wow... sounds like accountability is in short supply. I work in I.T., it is not my job to tell management how to do theirs. If I was an electrician, and I told management they shouldn't plug more than 30A into a 30A outlet, and they did it anyway and then the breaker tripped, it wouldn't be my fault. And if they somehow caught the building on fire, I would still sleep well at night. I like to think I work together with management to analyze complex scenarios, explain them in simple terms, recommend paths for improvement, and let management decide what to do. Granted, every work environment is different. But if I had to save copies of CYA emails so I didn't lose my job over poor management decisions, that would be my "resume generating event." -Jonathan -Original Message- From: Lightner, Jeff [mailto:jlight...@water.com] Sent: Thursday, September 29, 2011 11:47 AM To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Question on DB online backups (Wayne T Smith) Often enough "marginally important" becomes "critically important" only after it fails. Usually that is when the admin faces the firing squad because management is never going to admit THEY didn't see it as important before the failure. If it is important enough to backup then it is important enough to test the restore. Telling folks later that it didn't seem important to you is NOT going to be the correct answer especially if they instructed you to do the backup in the first place. The best you can do is to write an email to your management explaining the issues and keep a copy of it (and better yet any response) for CYA. Even then it doesn't always help. In one environment we repeatedly told management about lack of backup being an issue and they repeatedly let us know that they didn't see it as an issue. When it finally failed the response we got back from on high was "You weren't forceful enough in saying it was important." -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Martin, Jonathan Sent: Thursday, September 29, 2011 10:33 AM To: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) While I agree that someone could face the firing squad if a certain database backed critical application failed and there was no restore capability, this may not be *that* application. Let's face it, if this application isn't critical enough to bring in a DBA, even as a contractor, and they didn't pay for implementation services with backups, then this application may just be marginally important. I would come up with and test some sort of restore scenario and set your management's expectations. That scenario does not need to be 24x7 snapshot based point in time restore with tape backup to provide recovery. Perhaps you take down the DB once a week to do a full, and the expectation is that if the database goes down you can lose up to a weeks' worth of data. Perhaps you supplement with snapshots, or database exports. Either way, Wayne's write-up is a great place to start. -Jonathan -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of David McMullin Sent: Thursday, September 29, 2011 9:57 AM To: veritas-bu@mailman.eng.auburn.edu Cc: wtsm...@maine.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) Wayne - You are looking at this from the wrong perspective. You need to be concerned about RESTORING your data. It does not matter how 'successful' your backups are if you cannot restore the data. IMHO you really need to get a DBA or 'someone' to sign off on your backup procedure, as well as test your restore, otherwise you are setting yourself up for a "resume generating event"... -- Date: Tue, 27 Sep 2011 11:45:35 -0400 From: Wayne T Smith Subject: Re: [Veritas-bu] Question on DB online backups To: verita
Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
I respectfully disagree in all respects. - Refusing to get a database backed up until management hires a DBA could be a "resume generating event". - Like it or not, the questioner appears to be the Oracle DBA, albeit one very early in his DBA career and learning on his own! - [redacted] - Someone using NetBackup for as long as the questioner knows that a successful backup is not the same as being able to restore or meet expectations for recovery. - My post started with and ended with, essentially, "you need a DBA". - My intended perspective was that if management leaves the DBA job to the questioner, then a little high level knowledge will let him focus on getting the database protected at an appropriate level as quickly as possible. - What I wrote may be right or wrong, the perspective may be right or wrong for various circumstances, but for how I read this circumstance, the comment is off-base and helpful only in giving me pause before again "helping" in this forum. If this was the commenter's purpose, it worked. - If the commenter wrote "Wayne" when he meant the questioner ... never mind. Wayne NetBackup administrator Oracle database administrator thin-skinned today, apparently On Thu, Sep 29, 2011 at 9:56 AM, David McMullin < david.mcmul...@cbc-companies.com> wrote, in part: > Wayne - You are looking at this from the wrong perspective. > > You need to be concerned about RESTORING your data. > It does not matter how 'successful' your backups are if you cannot restore > the data. > > IMHO you really need to get a DBA or 'someone' to sign off on your backup > procedure, as well as test your restore, otherwise you are setting yourself > up for a "resume generating event"... > ... > ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
Wow... sounds like accountability is in short supply. I work in I.T., it is not my job to tell management how to do theirs. If I was an electrician, and I told management they shouldn't plug more than 30A into a 30A outlet, and they did it anyway and then the breaker tripped, it wouldn't be my fault. And if they somehow caught the building on fire, I would still sleep well at night. I like to think I work together with management to analyze complex scenarios, explain them in simple terms, recommend paths for improvement, and let management decide what to do. Granted, every work environment is different. But if I had to save copies of CYA emails so I didn't lose my job over poor management decisions, that would be my "resume generating event." -Jonathan -Original Message- From: Lightner, Jeff [mailto:jlight...@water.com] Sent: Thursday, September 29, 2011 11:47 AM To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Question on DB online backups (Wayne T Smith) Often enough "marginally important" becomes "critically important" only after it fails. Usually that is when the admin faces the firing squad because management is never going to admit THEY didn't see it as important before the failure. If it is important enough to backup then it is important enough to test the restore. Telling folks later that it didn't seem important to you is NOT going to be the correct answer especially if they instructed you to do the backup in the first place. The best you can do is to write an email to your management explaining the issues and keep a copy of it (and better yet any response) for CYA. Even then it doesn't always help. In one environment we repeatedly told management about lack of backup being an issue and they repeatedly let us know that they didn't see it as an issue. When it finally failed the response we got back from on high was "You weren't forceful enough in saying it was important." -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Martin, Jonathan Sent: Thursday, September 29, 2011 10:33 AM To: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) While I agree that someone could face the firing squad if a certain database backed critical application failed and there was no restore capability, this may not be *that* application. Let's face it, if this application isn't critical enough to bring in a DBA, even as a contractor, and they didn't pay for implementation services with backups, then this application may just be marginally important. I would come up with and test some sort of restore scenario and set your management's expectations. That scenario does not need to be 24x7 snapshot based point in time restore with tape backup to provide recovery. Perhaps you take down the DB once a week to do a full, and the expectation is that if the database goes down you can lose up to a weeks' worth of data. Perhaps you supplement with snapshots, or database exports. Either way, Wayne's write-up is a great place to start. -Jonathan -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of David McMullin Sent: Thursday, September 29, 2011 9:57 AM To: veritas-bu@mailman.eng.auburn.edu Cc: wtsm...@maine.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) Wayne - You are looking at this from the wrong perspective. You need to be concerned about RESTORING your data. It does not matter how 'successful' your backups are if you cannot restore the data. IMHO you really need to get a DBA or 'someone' to sign off on your backup procedure, as well as test your restore, otherwise you are setting yourself up for a "resume generating event"... -- Date: Tue, 27 Sep 2011 11:45:35 -0400 From: Wayne T Smith Subject: Re: [Veritas-bu] Question on DB online backups To: veritas-bu@mailman.eng.auburn.edu Message-ID: Content-Type: text/plain; charset="iso-8859-1" Ouch, having an Oracle database without a DBA is like having NetBackup without anyone that knows NetBackup. There are several ways to do backups of Oracle databases. These include - Cold, full. You take down the database and backup all associated disk space (data, redolog, and perhaps other types). - Cold backups do not require Oracle archivelog mode. - Restoration requires database down or restore to essentially identical setup on a like machine. - Restoration is to back to the time of your backup. - Note for all backup types: Your file system backups will exclude Oracle managed, as a file system backup of an online Oracle database is
Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
Often enough "marginally important" becomes "critically important" only after it fails. Usually that is when the admin faces the firing squad because management is never going to admit THEY didn't see it as important before the failure. If it is important enough to backup then it is important enough to test the restore. Telling folks later that it didn't seem important to you is NOT going to be the correct answer especially if they instructed you to do the backup in the first place. The best you can do is to write an email to your management explaining the issues and keep a copy of it (and better yet any response) for CYA. Even then it doesn't always help. In one environment we repeatedly told management about lack of backup being an issue and they repeatedly let us know that they didn't see it as an issue. When it finally failed the response we got back from on high was "You weren't forceful enough in saying it was important." -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Martin, Jonathan Sent: Thursday, September 29, 2011 10:33 AM To: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) While I agree that someone could face the firing squad if a certain database backed critical application failed and there was no restore capability, this may not be *that* application. Let's face it, if this application isn't critical enough to bring in a DBA, even as a contractor, and they didn't pay for implementation services with backups, then this application may just be marginally important. I would come up with and test some sort of restore scenario and set your management's expectations. That scenario does not need to be 24x7 snapshot based point in time restore with tape backup to provide recovery. Perhaps you take down the DB once a week to do a full, and the expectation is that if the database goes down you can lose up to a weeks' worth of data. Perhaps you supplement with snapshots, or database exports. Either way, Wayne's write-up is a great place to start. -Jonathan -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of David McMullin Sent: Thursday, September 29, 2011 9:57 AM To: veritas-bu@mailman.eng.auburn.edu Cc: wtsm...@maine.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) Wayne - You are looking at this from the wrong perspective. You need to be concerned about RESTORING your data. It does not matter how 'successful' your backups are if you cannot restore the data. IMHO you really need to get a DBA or 'someone' to sign off on your backup procedure, as well as test your restore, otherwise you are setting yourself up for a "resume generating event"... -- Date: Tue, 27 Sep 2011 11:45:35 -0400 From: Wayne T Smith Subject: Re: [Veritas-bu] Question on DB online backups To: veritas-bu@mailman.eng.auburn.edu Message-ID: Content-Type: text/plain; charset="iso-8859-1" Ouch, having an Oracle database without a DBA is like having NetBackup without anyone that knows NetBackup. There are several ways to do backups of Oracle databases. These include - Cold, full. You take down the database and backup all associated disk space (data, redolog, and perhaps other types). - Cold backups do not require Oracle archivelog mode. - Restoration requires database down or restore to essentially identical setup on a like machine. - Restoration is to back to the time of your backup. - Note for all backup types: Your file system backups will exclude Oracle managed, as a file system backup of an online Oracle database is insufficient for recovery. Your file system backup should include the Oracle software home and certain other objects (control files, inventory, oraInst.loc, etc) ... not sure where these are on Windows. - Cold, RMAN level 0 and 1. You write a script that brings the database down, use RMAN to back it up, then start the database again. - Note for all RMAN backup types: RMAN is simply the Oracle utility to do backup and restore. - Whereas the above backups were simply of file system disk spaces while the database is down (perhaps using NetBackup or a disk copy utility), RMAN decides what data to copy, where to put it and keeps track where it has put the backup files. - RMAN writes its backups to disk or tape. While it is possible to have RMAN write to disk and then have NetBackup backup the file system data, this is awkward and restoration goes from a simple, automatic process to a time-consuming very difficult process if NetB
Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
While I agree that someone could face the firing squad if a certain database backed critical application failed and there was no restore capability, this may not be *that* application. Let's face it, if this application isn't critical enough to bring in a DBA, even as a contractor, and they didn't pay for implementation services with backups, then this application may just be marginally important. I would come up with and test some sort of restore scenario and set your management's expectations. That scenario does not need to be 24x7 snapshot based point in time restore with tape backup to provide recovery. Perhaps you take down the DB once a week to do a full, and the expectation is that if the database goes down you can lose up to a weeks' worth of data. Perhaps you supplement with snapshots, or database exports. Either way, Wayne's write-up is a great place to start. -Jonathan -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of David McMullin Sent: Thursday, September 29, 2011 9:57 AM To: veritas-bu@mailman.eng.auburn.edu Cc: wtsm...@maine.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) Wayne - You are looking at this from the wrong perspective. You need to be concerned about RESTORING your data. It does not matter how 'successful' your backups are if you cannot restore the data. IMHO you really need to get a DBA or 'someone' to sign off on your backup procedure, as well as test your restore, otherwise you are setting yourself up for a "resume generating event"... -- Date: Tue, 27 Sep 2011 11:45:35 -0400 From: Wayne T Smith Subject: Re: [Veritas-bu] Question on DB online backups To: veritas-bu@mailman.eng.auburn.edu Message-ID: Content-Type: text/plain; charset="iso-8859-1" Ouch, having an Oracle database without a DBA is like having NetBackup without anyone that knows NetBackup. There are several ways to do backups of Oracle databases. These include - Cold, full. You take down the database and backup all associated disk space (data, redolog, and perhaps other types). - Cold backups do not require Oracle archivelog mode. - Restoration requires database down or restore to essentially identical setup on a like machine. - Restoration is to back to the time of your backup. - Note for all backup types: Your file system backups will exclude Oracle managed, as a file system backup of an online Oracle database is insufficient for recovery. Your file system backup should include the Oracle software home and certain other objects (control files, inventory, oraInst.loc, etc) ... not sure where these are on Windows. - Cold, RMAN level 0 and 1. You write a script that brings the database down, use RMAN to back it up, then start the database again. - Note for all RMAN backup types: RMAN is simply the Oracle utility to do backup and restore. - Whereas the above backups were simply of file system disk spaces while the database is down (perhaps using NetBackup or a disk copy utility), RMAN decides what data to copy, where to put it and keeps track where it has put the backup files. - RMAN writes its backups to disk or tape. While it is possible to have RMAN write to disk and then have NetBackup backup the file system data, this is awkward and restoration goes from a simple, automatic process to a time-consuming very difficult process if NetBackup has the data. - The Netbackup solution is to purchase a license that includes the Oracle Agent. This is a shim that gets installed in the Oracle software home. RMAN thinks it is backing up to tape (device type "sbt_tape"), but the shim captures the RMAN data and sends it on to your backup server. It doesn't matter if your backup server uses disk or tape ... everything back at the backup server is transparent to RMAN ... just like NetBackup file system backups. - Just like file system cold backups, you need to verify and practice various restore scenarios. RMAN gives you a much better chance to do the restore you need (and have the necessary backup objects available). - RMAN keeps track of the stuff it backs up. It has two methods. RMAN will put information about its backups in the Oracle database control files. RMAN also has a "catalog" feature, which means its backup information is stored in a database someplace. If you use the RMAN catalog, your restore scenarios are substantially enhanced. Using an RMAN catalog is NOT required by RMAN nor the Oracle Agent. - RMAN has its own retention schemes. Now you have 2
Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
I always tell people my DR plan is to keep an updated resume handy. :-) -Original Message- From: veritas-bu-boun...@mailman.eng.auburn.edu [mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of David McMullin Sent: Thursday, September 29, 2011 9:57 AM To: veritas-bu@mailman.eng.auburn.edu Cc: wtsm...@maine.edu Subject: Re: [Veritas-bu] Question on DB online backups (Wayne T Smith) Wayne - You are looking at this from the wrong perspective. You need to be concerned about RESTORING your data. It does not matter how 'successful' your backups are if you cannot restore the data. IMHO you really need to get a DBA or 'someone' to sign off on your backup procedure, as well as test your restore, otherwise you are setting yourself up for a "resume generating event"... -- Date: Tue, 27 Sep 2011 11:45:35 -0400 From: Wayne T Smith Subject: Re: [Veritas-bu] Question on DB online backups To: veritas-bu@mailman.eng.auburn.edu Message-ID: Content-Type: text/plain; charset="iso-8859-1" Ouch, having an Oracle database without a DBA is like having NetBackup without anyone that knows NetBackup. There are several ways to do backups of Oracle databases. These include - Cold, full. You take down the database and backup all associated disk space (data, redolog, and perhaps other types). - Cold backups do not require Oracle archivelog mode. - Restoration requires database down or restore to essentially identical setup on a like machine. - Restoration is to back to the time of your backup. - Note for all backup types: Your file system backups will exclude Oracle managed, as a file system backup of an online Oracle database is insufficient for recovery. Your file system backup should include the Oracle software home and certain other objects (control files, inventory, oraInst.loc, etc) ... not sure where these are on Windows. - Cold, RMAN level 0 and 1. You write a script that brings the database down, use RMAN to back it up, then start the database again. - Note for all RMAN backup types: RMAN is simply the Oracle utility to do backup and restore. - Whereas the above backups were simply of file system disk spaces while the database is down (perhaps using NetBackup or a disk copy utility), RMAN decides what data to copy, where to put it and keeps track where it has put the backup files. - RMAN writes its backups to disk or tape. While it is possible to have RMAN write to disk and then have NetBackup backup the file system data, this is awkward and restoration goes from a simple, automatic process to a time-consuming very difficult process if NetBackup has the data. - The Netbackup solution is to purchase a license that includes the Oracle Agent. This is a shim that gets installed in the Oracle software home. RMAN thinks it is backing up to tape (device type "sbt_tape"), but the shim captures the RMAN data and sends it on to your backup server. It doesn't matter if your backup server uses disk or tape ... everything back at the backup server is transparent to RMAN ... just like NetBackup file system backups. - Just like file system cold backups, you need to verify and practice various restore scenarios. RMAN gives you a much better chance to do the restore you need (and have the necessary backup objects available). - RMAN keeps track of the stuff it backs up. It has two methods. RMAN will put information about its backups in the Oracle database control files. RMAN also has a "catalog" feature, which means its backup information is stored in a database someplace. If you use the RMAN catalog, your restore scenarios are substantially enhanced. Using an RMAN catalog is NOT required by RMAN nor the Oracle Agent. - RMAN has its own retention schemes. Now you have 2 retentions to worry about ... if either the RMAN retention or the NetBackup retention period expires, your backup data is lost. - Hot - Hot backups are taken with the database online. Hot backups require archivelog mode set in the database, which means that changes to the database, as recorded in the redologs that any Oracle database has, are copied to archive redo logs. Archivelog mode along with database backups allow one to restore/recover to a point in time of your choice (that is, all committed changes at any point in time). Depending on your requirements, these archive redo logs must be saved for as far back as you might wish to do a restore. - Hot backups do not require RMAN. One may put an Oracle tablespace (collection of related data files) in
Re: [Veritas-bu] Question on DB online backups (Wayne T Smith)
Wayne - You are looking at this from the wrong perspective. You need to be concerned about RESTORING your data. It does not matter how 'successful' your backups are if you cannot restore the data. IMHO you really need to get a DBA or 'someone' to sign off on your backup procedure, as well as test your restore, otherwise you are setting yourself up for a "resume generating event"... -- Date: Tue, 27 Sep 2011 11:45:35 -0400 From: Wayne T Smith Subject: Re: [Veritas-bu] Question on DB online backups To: veritas-bu@mailman.eng.auburn.edu Message-ID: Content-Type: text/plain; charset="iso-8859-1" Ouch, having an Oracle database without a DBA is like having NetBackup without anyone that knows NetBackup. There are several ways to do backups of Oracle databases. These include - Cold, full. You take down the database and backup all associated disk space (data, redolog, and perhaps other types). - Cold backups do not require Oracle archivelog mode. - Restoration requires database down or restore to essentially identical setup on a like machine. - Restoration is to back to the time of your backup. - Note for all backup types: Your file system backups will exclude Oracle managed, as a file system backup of an online Oracle database is insufficient for recovery. Your file system backup should include the Oracle software home and certain other objects (control files, inventory, oraInst.loc, etc) ... not sure where these are on Windows. - Cold, RMAN level 0 and 1. You write a script that brings the database down, use RMAN to back it up, then start the database again. - Note for all RMAN backup types: RMAN is simply the Oracle utility to do backup and restore. - Whereas the above backups were simply of file system disk spaces while the database is down (perhaps using NetBackup or a disk copy utility), RMAN decides what data to copy, where to put it and keeps track where it has put the backup files. - RMAN writes its backups to disk or tape. While it is possible to have RMAN write to disk and then have NetBackup backup the file system data, this is awkward and restoration goes from a simple, automatic process to a time-consuming very difficult process if NetBackup has the data. - The Netbackup solution is to purchase a license that includes the Oracle Agent. This is a shim that gets installed in the Oracle software home. RMAN thinks it is backing up to tape (device type "sbt_tape"), but the shim captures the RMAN data and sends it on to your backup server. It doesn't matter if your backup server uses disk or tape ... everything back at the backup server is transparent to RMAN ... just like NetBackup file system backups. - Just like file system cold backups, you need to verify and practice various restore scenarios. RMAN gives you a much better chance to do the restore you need (and have the necessary backup objects available). - RMAN keeps track of the stuff it backs up. It has two methods. RMAN will put information about its backups in the Oracle database control files. RMAN also has a "catalog" feature, which means its backup information is stored in a database someplace. If you use the RMAN catalog, your restore scenarios are substantially enhanced. Using an RMAN catalog is NOT required by RMAN nor the Oracle Agent. - RMAN has its own retention schemes. Now you have 2 retentions to worry about ... if either the RMAN retention or the NetBackup retention period expires, your backup data is lost. - Hot - Hot backups are taken with the database online. Hot backups require archivelog mode set in the database, which means that changes to the database, as recorded in the redologs that any Oracle database has, are copied to archive redo logs. Archivelog mode along with database backups allow one to restore/recover to a point in time of your choice (that is, all committed changes at any point in time). Depending on your requirements, these archive redo logs must be saved for as far back as you might wish to do a restore. - Hot backups do not require RMAN. One may put an Oracle tablespace (collection of related data files) in backup mode, backup (by disk copy, NetBackup user backup, or whatever) and then remove backup mode. While this can be done I strongly suggest you don't, for I predict you will not be able to do the restore you want one day. - RMAN is the tool of choice for hot backups. Again, backup may be to RMAN disk or tape. - I backup many databases using the NetBackup Oracle Agent. With an appropriately written scri