{"id":776,"date":"2020-07-27T17:10:36","date_gmt":"2020-07-27T20:10:36","guid":{"rendered":"https:\/\/www.fernandosimon.com\/blog\/?p=776"},"modified":"2020-07-27T17:10:36","modified_gmt":"2020-07-27T20:10:36","slug":"zdlra-protecting-databases-with-replication","status":"publish","type":"post","link":"https:\/\/www.fernandosimon.com\/blog\/zdlra-protecting-databases-with-replication\/","title":{"rendered":"ZDLRA, Protecting Databases with Replication"},"content":{"rendered":"

The replication for ZDLRA works differently than normal DataGuard, but you can reach almost the same level of multiple site protection with that. The replication for ZDLRA is not complicated but can be divided into several steps. Basically, to protect a database (since you have everything configure) is done linking the database with the protection policy that is replicated.<\/p>\n

In my previous posts, I already wrote about all the steps to reach this configuration. Starting with an explanation about replication<\/a>, how to configure the replication network<\/a> between ZDLRA\u2019s, how to configure the replication server<\/a>,  and how to create the replication config<\/a> (that links everything is done before).<\/p>\n

\"\"<\/a><\/p>\n

But most of the time we don\u2019t need to pass through all of these steps. Usually, the ZDLRA is deployed with the replication network already configured, or you already deploy two ZDLRA\u2019s that will operate replicated. This part I consider the \u201cphysical\u201d part of the configuration because evolves network and details that we usually don\u2019t touch after configured. The \u201clogical\u201d part comes after and evolves all the definitions about what policies will be replicated, which databases will be part of each policy, and so on. This \u201clogical\u201d configuration I explained in this previous post<\/a>.<\/p>\n

But it is important to know how it\u2019s working to understand all the details. And if you need, you can check my posts about the replication for ZDLRA<\/a>.<\/p>\n

In this post, I will show more details on how the ZDLRA replication impact over the backup for your database, and show the protection occurs to reduce your RPO to a minimum.<\/p>\n

<\/p>\n

Replication, Policies, and Databases<\/h2>\n

A little recap. The most important to understand about the replication is that is done based on protection policies. ZDLRA does not replicate database per database, it replicates policies. So, every database that is linked with the policy will be replicated.<\/p>\n

You can have more than one protection policy that is replicated between the same ZDLRA\u2019s, and this link directly with the MAA Architecture where you have more than one SLA that needs to be sustained (like recovery window requirements as an example). And the caveat it is that, maybe, you need to double part of your policies because it is possible that you don\u2019t want to replicate all databases for that policy. So, you will create another policy and drop all databases to be replicated there. But remember that when you do that, you need to check again your clone to tapes because they can be linked with policies (and not with backups tags of another configuration). And if this occurs, you will need to reorganize your clone to the tape schedule.<\/p>\n

It is important to also understand that replication is done as soon as possible. So, after every backup that is ingested at upstream ZDLRA (and after it is validated), it is replicated ASAP. I already explained this in previous posts, but the full backup it is not replicated, just the incremental and the archivelogs are. Think that your datafile has more than 20TB, you will wait until the virtual full is generated and replicate after that? If you do this you lose protection because, in case of an entire datacenter failure, your backup is not replicate in a secure environment.  And since the virtual full backup will be created (at downstream) it will be equal to that one that was created at upstream.<\/p>\n

Base Config<\/h2>\n

The base config is the same as I create in my previous post<\/a>. I have the ORCL19 database that is linked with the protection policy ZDLRAUP_BRONZE trough a replication server called ZDLRAS2_REP.<\/p>\n

Replicated Protection<\/h2>\n

To use the replication we just need to do a backup of the database. So, below you can see a simple backup for datafile 1. This backup is the first backup that I am doing for the database at this ZDLRA, so it is a level 0 backup.<\/p>\n

First, look that I connected at upstream ZDLRA (ZDLRAS1 in this scenario).<\/p>\n

[oracle@orcloel7 ~]$ rman target=\/ catalog=vpcsrc\/vpcsrc@zdlras1-scan:1521\/zdlras1\r\n\r\nRecovery Manager: Release 19.0.0.0.0 - Production on Thu Jul 9 22:54:55 2020\r\nVersion 19.3.0.0.0\r\n\r\nCopyright (c) 1982, 2019, Oracle and\/or its affiliates.  All rights reserved.\r\n\r\nconnected to target database: ORCL19 (DBID=323177095)\r\nconnected to recovery catalog database\r\n\r\nRMAN><\/pre>\n
And executed the first backup of datafile 1:<\/p>\n
RMAN> BACKUP INCREMENTAL LEVEL 0 DEVICE TYPE SBT FILESPERSET 1 DATAFILE 1 TAG 'BKP-DB-INCR0-REP';\r\n\r\nStarting backup at 09\/07\/2020 23:31:06\r\nreleased channel: ORA_DISK_1\r\nallocated channel: ORA_SBT_TAPE_1\r\nchannel ORA_SBT_TAPE_1: SID=37 device type=SBT_TAPE\r\nchannel ORA_SBT_TAPE_1: RA Library (ZDLRAS1) SID=AA098857038E149DE053010310AC801E\r\nchannel ORA_SBT_TAPE_1: starting incremental level 0 datafile backup set\r\nchannel ORA_SBT_TAPE_1: specifying datafile(s) in backup set\r\ninput datafile file number=00001 name=\/u01\/app\/oracle\/oradata\/ORCL19\/system01.dbf\r\nchannel ORA_SBT_TAPE_1: starting piece 1 at 09\/07\/2020 23:31:09\r\nchannel ORA_SBT_TAPE_1: finished piece 1 at 09\/07\/2020 23:33:24\r\npiece handle=ORCL19_3fv4tjdt_1_1 tag=BKP-DB-INCR0-REP comment=API Version 2.0,MMS Version 12.2.0.2\r\nchannel ORA_SBT_TAPE_1: backup set complete, elapsed time: 00:02:15\r\nFinished backup at 09\/07\/2020 23:33:24\r\n\r\nStarting Control File and SPFILE Autobackup at 09\/07\/2020 23:33:24\r\npiece handle=c-323177095-20200709-02 comment=API Version 2.0,MMS Version 12.2.0.2\r\nFinished Control File and SPFILE Autobackup at 09\/07\/2020 23:33:41\r\n\r\nRMAN><\/pre>\n
And as you can see here, it was replicated:<\/p>\n
RMAN> list backup of datafile 1 tag = 'BKP-DB-INCR0-REP';\r\n\r\n\r\nList of Backup Sets\r\n===================\r\n\r\n\r\nBS Key  Type LV Size\r\n------- ---- -- ----------\r\n25797   Incr 0  325.02M\r\n  List of Datafiles in backup set 25797\r\n  File LV Type Ckp SCN    Ckp Time            Abs Fuz SCN Sparse Name\r\n  ---- -- ---- ---------- ------------------- ----------- ------ ----\r\n  1    0  Incr 2414591    09\/07\/2020 23:31:09              NO    \/u01\/app\/oracle\/oradata\/ORCL19\/system01.dbf\r\n\r\n  Backup Set Copy #1 of backup set 25797\r\n  Device Type Elapsed Time Completion Time     Compressed Tag\r\n  ----------- ------------ ------------------- ---------- ---\r\n  SBT_TAPE    00:04:57     09\/07\/2020 23:36:11 YES        BKP-DB-INCR0-REP\r\n\r\n    List of Backup Pieces for backup set 25797 Copy #1\r\n    BP Key  Pc# Status      Media                   Piece Name\r\n    ------- --- ----------- ----------------------- ----------\r\n    25798   1   AVAILABLE                           VB$_1891149551_25782I\r\n\r\n  Backup Set Copy #2 of backup set 25797\r\n  Device Type Elapsed Time Completion Time     Compressed Tag\r\n  ----------- ------------ ------------------- ---------- ---\r\n  SBT_TAPE    00:04:57     09\/07\/2020 23:36:06 YES        BKP-DB-INCR0-REP\r\n\r\n    List of Backup Pieces for backup set 25797 Copy #2\r\n    BP Key  Pc# Status      Media                   Piece Name\r\n    ------- --- ----------- ----------------------- ----------\r\n    25811   1   AVAILABLE   ZDLRAS2_REP             VB$_2127575003_12535I\r\n\r\nRMAN><\/pre>\n
Some details here:<\/p>\n