snapshots of db disks on SAN

IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com

I wouldn't trust it. At best it would only be able to recover reliably to the last checkpoint with no rollforward/rollback. At worst there would be inconsistencies that the server could not resolve and it might not come online at all. This is a job for HDR not disk based replication. Art S. Kagel ----- Original Message ----- From: Robert Roussey <ids@iiug.org> At: 10/19 10:24:23 IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com ******************************************************************************* Forum Note: Use "Reply" to post a response in the discussion forum.

You will need to read the manual about "external backups" You have the ability to use a blocking check point "onmode -c block" and once the snap is complete you issue the "onmode -c unblock" This will let the DBSEREVER know there is an external backup (there are directions to do external backup and restore in the manuals. We use the external backup to bring up a secondary server (for testing and read only purposes on a very large Operation Data Store. (not on 10.00 though on an earlier release) We also archive the secondary to tape (the secondary is not part of HDR or CDR and gets overwritten daily and would the company would like the ability to restore to a given data and time from more then a day ago if we need to. Local logs can be replayed as part of the "external" restore. Also onstat -g arc shows when the last external (or regular) archive has been taken. NOTE of importance. If the snap is not completed before the next check point there will be a period of time when no writes/updates will be allowed. The lock up will be equivalent to a very long checkpoint. George -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of ROBERT ROUSSEY Sent: Thursday, October 19, 2006 7:17 AM To: ids@iiug.org Subject: snapshots of db disks on SAN [7648] IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum.

We use the EMC with Informix9.4.fc5 on Solaris and sucessfully use mirror copies. We sync up the mirrors between two EMC units and once in sync we issue the onmode -g block then break the connection then issue the onmode -c unblock. We have used thes copies in test but also as our offsite DR point. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of George Palmer Sent: Thursday, October 19, 2006 9:45 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7650] You will need to read the manual about "external backups" You have the ability to use a blocking check point "onmode -c block" and once the snap is complete you issue the "onmode -c unblock" This will let the DBSEREVER know there is an external backup (there are directions to do external backup and restore in the manuals. We use the external backup to bring up a secondary server (for testing and read only purposes on a very large Operation Data Store. (not on 10.00 though on an earlier release) We also archive the secondary to tape (the secondary is not part of HDR or CDR and gets overwritten daily and would the company would like the ability to restore to a given data and time from more then a day ago if we need to. Local logs can be replayed as part of the "external" restore. Also onstat -g arc shows when the last external (or regular) archive has been taken. NOTE of importance. If the snap is not completed before the next check point there will be a period of time when no writes/updates will be allowed. The lock up will be equivalent to a very long checkpoint. George -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of ROBERT ROUSSEY Sent: Thursday, October 19, 2006 7:17 AM To: ids@iiug.org Subject: snapshots of db disks on SAN [7648] IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum.

How long do you have to do the block? Bob Roussey Robert.Roussey@SpiritAir.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Bruce Simms Sent: Thursday, October 19, 2006 11:09 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7651] We use the EMC with Informix9.4.fc5 on Solaris and sucessfully use mirror copies. We sync up the mirrors between two EMC units and once in sync we issue the onmode -g block then break the connection then issue the onmode -c unblock. We have used thes copies in test but also as our offsite DR point. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of George Palmer Sent: Thursday, October 19, 2006 9:45 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7650] You will need to read the manual about "external backups" You have the ability to use a blocking check point "onmode -c block" and once the snap is complete you issue the "onmode -c unblock" This will let the DBSEREVER know there is an external backup (there are directions to do external backup and restore in the manuals. We use the external backup to bring up a secondary server (for testing and read only purposes on a very large Operation Data Store. (not on 10.00 though on an earlier release) We also archive the secondary to tape (the secondary is not part of HDR or CDR and gets overwritten daily and would the company would like the ability to restore to a given data and time from more then a day ago if we need to. Local logs can be replayed as part of the "external" restore. Also onstat -g arc shows when the last external (or regular) archive has been taken. NOTE of importance. If the snap is not completed before the next check point there will be a period of time when no writes/updates will be allowed. The lock up will be equivalent to a very long checkpoint. George -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of ROBERT ROUSSEY Sent: Thursday, October 19, 2006 7:17 AM To: ids@iiug.org Subject: snapshots of db disks on SAN [7648] IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum.

Bob, We issue the block in a shell upon the return from the block we issue the breaking of the mirror followed by the unblock. This whole action takes less tan 4 seconds ton our systems. *** Please note the new address and phone number below. Bruce Simms Data Base Services TALX Corporation 1850 Borman Court St. Louis, MO 63146 Phone (314) 214-7703 FAX (314) 983-3238 bsimms@talx.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Robert Roussey(MIS) Sent: Monday, October 23, 2006 9:35 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7662] How long do you have to do the block? Bob Roussey Robert.Roussey@SpiritAir.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Bruce Simms Sent: Thursday, October 19, 2006 11:09 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7651] We use the EMC with Informix9.4.fc5 on Solaris and sucessfully use mirror copies. We sync up the mirrors between two EMC units and once in sync we issue the onmode -g block then break the connection then issue the onmode -c unblock. We have used thes copies in test but also as our offsite DR point. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of George Palmer Sent: Thursday, October 19, 2006 9:45 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7650] You will need to read the manual about "external backups" You have the ability to use a blocking check point "onmode -c block" and once the snap is complete you issue the "onmode -c unblock" This will let the DBSEREVER know there is an external backup (there are directions to do external backup and restore in the manuals. We use the external backup to bring up a secondary server (for testing and read only purposes on a very large Operation Data Store. (not on 10.00 though on an earlier release) We also archive the secondary to tape (the secondary is not part of HDR or CDR and gets overwritten daily and would the company would like the ability to restore to a given data and time from more then a day ago if we need to. Local logs can be replayed as part of the "external" restore. Also onstat -g arc shows when the last external (or regular) archive has been taken. NOTE of importance. If the snap is not completed before the next check point there will be a period of time when no writes/updates will be allowed. The lock up will be equivalent to a very long checkpoint. George -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of ROBERT ROUSSEY Sent: Thursday, October 19, 2006 7:17 AM To: ids@iiug.org Subject: snapshots of db disks on SAN [7648] IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum.

The length of time for the 'block' of the instance varies, you'll need trial and error to discover the optimal result. You'll find the more LUNs the instance uses the longer the time: In addition factors like the SAN flare code versions and logging mode of the database all play their parts. I quite envy Bruce, who can get the round trip done in less than four seconds :) mine take around twelve seconds. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Bruce Simms Sent: Tuesday, 24 October 2006 00:38 To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7663] Bob, We issue the block in a shell upon the return from the block we issue the breaking of the mirror followed by the unblock. This whole action takes less tan 4 seconds ton our systems. *** Please note the new address and phone number below. Bruce Simms Data Base Services TALX Corporation 1850 Borman Court St. Louis, MO 63146 Phone (314) 214-7703 FAX (314) 983-3238 bsimms@talx.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Robert Roussey(MIS) Sent: Monday, October 23, 2006 9:35 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7662] How long do you have to do the block? Bob Roussey Robert.Roussey@SpiritAir.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Bruce Simms Sent: Thursday, October 19, 2006 11:09 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7651] We use the EMC with Informix9.4.fc5 on Solaris and sucessfully use mirror copies. We sync up the mirrors between two EMC units and once in sync we issue the onmode -g block then break the connection then issue the onmode -c unblock. We have used thes copies in test but also as our offsite DR point. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of George Palmer Sent: Thursday, October 19, 2006 9:45 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7650] You will need to read the manual about "external backups" You have the ability to use a blocking check point "onmode -c block" and once the snap is complete you issue the "onmode -c unblock" This will let the DBSEREVER know there is an external backup (there are directions to do external backup and restore in the manuals. We use the external backup to bring up a secondary server (for testing and read only purposes on a very large Operation Data Store. (not on 10.00 though on an earlier release) We also archive the secondary to tape (the secondary is not part of HDR or CDR and gets overwritten daily and would the company would like the ability to restore to a given data and time from more then a day ago if we need to. Local logs can be replayed as part of the "external" restore. Also onstat -g arc shows when the last external (or regular) archive has been taken. NOTE of importance. If the snap is not completed before the next check point there will be a period of time when no writes/updates will be allowed. The lock up will be equivalent to a very long checkpoint. George -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of ROBERT ROUSSEY Sent: Thursday, October 19, 2006 7:17 AM To: ids@iiug.org Subject: snapshots of db disks on SAN [7648] IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ******************************************************************* Everything about this email and its attachments, including potentially confidential components, is only for the eyes and ears of the persons to whom it has been addressed (i.e. the persons whose names appear in the "To" section of the email; however, those listed in sections "CC" and "BCC" may also consider themselves included in the group whose eyes and ears this email is for.). In the event that no physical, emotional or spiritual resemblance can be made between you and the intended recipients, you have been mistakenly or deliberately omitted from the email, someone has given it to you, or you have nicked it. If you're not supposed to have access to this email, please delete it, destroy it and deprive others from it. Oh, and let us know when you are done. *******************************************************************

Please keep in mind under the EMC product we have issued the sync command to the EMC system at a prior time so the mirroring might take 30 to 40 minutes to re establish the mirrors and sync them while we continue business as usual we only issue the block/unblock while we are breaking the mirrors this is why it is such a short time. *** Please note the new address and phone number below. Bruce Simms Data Base Services TALX Corporation 1850 Borman Court St. Louis, MO 63146 Phone (314) 214-7703 FAX (314) 983-3238 bsimms@talx.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Clifford Burton Sent: Monday, October 23, 2006 5:30 PM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7665] The length of time for the 'block' of the instance varies, you'll need trial and error to discover the optimal result. You'll find the more LUNs the instance uses the longer the time: In addition factors like the SAN flare code versions and logging mode of the database all play their parts. I quite envy Bruce, who can get the round trip done in less than four seconds :) mine take around twelve seconds. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Bruce Simms Sent: Tuesday, 24 October 2006 00:38 To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7663] Bob, We issue the block in a shell upon the return from the block we issue the breaking of the mirror followed by the unblock. This whole action takes less tan 4 seconds ton our systems. *** Please note the new address and phone number below. Bruce Simms Data Base Services TALX Corporation 1850 Borman Court St. Louis, MO 63146 Phone (314) 214-7703 FAX (314) 983-3238 bsimms@talx.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Robert Roussey(MIS) Sent: Monday, October 23, 2006 9:35 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7662] How long do you have to do the block? Bob Roussey Robert.Roussey@SpiritAir.com -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Bruce Simms Sent: Thursday, October 19, 2006 11:09 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7651] We use the EMC with Informix9.4.fc5 on Solaris and sucessfully use mirror copies. We sync up the mirrors between two EMC units and once in sync we issue the onmode -g block then break the connection then issue the onmode -c unblock. We have used thes copies in test but also as our offsite DR point. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of George Palmer Sent: Thursday, October 19, 2006 9:45 AM To: ids@iiug.org Subject: RE: snapshots of db disks on SAN [7650] You will need to read the manual about "external backups" You have the ability to use a blocking check point "onmode -c block" and once the snap is complete you issue the "onmode -c unblock" This will let the DBSEREVER know there is an external backup (there are directions to do external backup and restore in the manuals. We use the external backup to bring up a secondary server (for testing and read only purposes on a very large Operation Data Store. (not on 10.00 though on an earlier release) We also archive the secondary to tape (the secondary is not part of HDR or CDR and gets overwritten daily and would the company would like the ability to restore to a given data and time from more then a day ago if we need to. Local logs can be replayed as part of the "external" restore. Also onstat -g arc shows when the last external (or regular) archive has been taken. NOTE of importance. If the snap is not completed before the next check point there will be a period of time when no writes/updates will be allowed. The lock up will be equivalent to a very long checkpoint. George -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of ROBERT ROUSSEY Sent: Thursday, October 19, 2006 7:17 AM To: ids@iiug.org Subject: snapshots of db disks on SAN [7648] IDS 10.00.UC4 / RedHat AS 3. We are moving to a configuration with all of our db disks on a SAN (Dell/EMC?). I was told that the SAN itself does 'snapshots' to other disks as often as we configure it. My mgmt spec'ing and ordering the equipment was told that these snapshots to a mirrored SAN should be fine to bring up the db if the primary SAN goes down. We currently use ontape for full backups. I'm concerned that the SAN snapshots don't take into consideration the checkpoint and timestamp info that ontape does. Does anyone else have their db disks on a SAN and done a successful failover / switch-over to a backup SAN *with* the db running with a full load; ie: without quiescing the db? TIA Bob Roussey robert.roussey@spiritair.com ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum. ******************************************************************* Everything about this email and its attachments, including potentially confidential components, is only for the eyes and ears of the persons to whom it has been addressed (i.e. the persons whose names appear in the "To" section of the email; however, those listed in sections "CC" and "BCC" may also consider themselves included in the group whose eyes and ears this email is for.). In the event that no physical, emotional or spiritual resemblance can be made between you and the intended recipients, you have been mistakenly or deliberately omitted from the email, someone has given it to you, or you have nicked it. If you're not supposed to have access to this email, please delete it, destroy it and deprive others from it. Oh, and let us know when you are done. ******************************************************************* ************************************************************************ ******* Forum Note: Use "Reply" to post a response in the discussion forum.