pat98's always & forever

19c 까지는 되었는데 21c 이후부터 Clusterware 와의 연동이 desupport 되었다.

공유영역으로 구성이 가능한지 확인해 보기 위해 테스트 진행해 보았다. 

사실 공식적으로 지원하지 않는다고 명시된 부분이라 절대 현업에는 적용하면 안될 것이다.

https://docs.oracle.com/en/database/oracle/oracle-database/23/upgrd/oracle-database-changes-deprecations-desupports.html#GUID-3AAE48F4-68CC-469C-B682-CBE521B19AEF

Desupport of Vendor Clusterware Integration with Oracle Clusterware

Starting with Oracle Clusterware 21c, the integration of vendor or third party clusterware with Oracle Clusterware is desupported.

Date: May 2021

The integration of vendor clusterware with Oracle Clusterware is desupported in Oracle Database 21c. Desupporting certain clustering features with limited adoption allows Oracle to focus on improving core scaling, availability, and manageability across all features and functionality. In the absence of an integration between different cluster solutions, the system is subject to the dueling cluster solutions issue: Independent cluster solutions can make individual decisions about which corrective actions must be taken in case of certain failures. To avoid conflicts, only one cluster solution should be active at any point in time. For this reason, Oracle recommends that you align your next software or hardware upgrade with the transition off of vendor cluster solutions. This desupport applies to clustered deployments, including Oracle Real Application Clusters and Oracle RAC One Node.

OS : Oracle Linux 9.7

공유 클러스터 : OCFS2

오라클 GI 버전  : 23.26.3.0.0
39578862;MICRONAUT RELEASE UPDATE 23.26.3.0.0 (39578862) Gold Image
39578856;RHP RELEASE UPDATE 23.26.3.0.0 (39578856) Gold Image
39578865;ACFS RELEASE UPDATE 23.26.3.0.0 (39578865) Gold Image
39578859;OCW RELEASE UPDATE 23.26.3.0.0 (39578859) Gold Image
39578879;Database Release Update : 23.26.3.0.0 (39578879) Gold Image

우선 테스트를 위해 OCFS 로 아래와 같이 공유영역을 구성해 놓았다.

[root@ocfs1 ~]# df -h
Filesystem                 Size  Used Avail Use% Mounted on
devtmpfs                   4.0M     0  4.0M   0% /dev
tmpfs                      4.9G     0  4.9G   0% /dev/shm
tmpfs                      2.0G  8.9M  2.0G   1% /run
/dev/mapper/ol_ocfs1-root   11G  5.5G  5.5G  51% /
/dev/sda1                  960M  427M  534M  45% /boot
/dev/mapper/ol_ocfs1-u01    40G  318M   40G   1% /u01
/dev/sdb                    20G  1.1G   19G   6% /oradata
/dev/sdc                    10G  551M  9.5G   6% /archive
/dev/sdd                   2.0G  279M  1.8G  14% /vote01
/dev/sdf                   2.0G  279M  1.8G  14% /vote02
/dev/sde                   2.0G  279M  1.8G  14% /vote03
tmpfs                      995M   36K  995M   1% /run/user/0

 GI SW 를 구성하기 위해서는  VOTE 디스크가 공유영역에 있어야 하는데 아래와 같이 위치를 지정하면 26ai 은 19c 와 달리 이렇게 에러를 뿜어낸다.

Cause - The storage type of the location (/vote01/ocr01) is not supported for Oracle Cluster Registry.  
Action - Provide a supported storage location for the Oracle Cluster Registry.  
Additional Information:
- PRVG-11503 : Path "/vote01" of type "OCFS2" is not suitable for usage as Oracle Clusterware storage for release "23.0". Supported storage types are "NFS, ASM Disk Group, VXFS"
- Cause: The specified path is not found suitable for usage as Oracle Clusterware storage(OCR or Voting Disk)
- Action: Ensure that you have selected a path that is suitable for the desired usage. Summary of node specific errors ocfs2 ocfs1 
 

안내메세지는 위와 같이 클러스터 저장 영역으로는 NFS, ASM Disk, VXFS (베리타스 파일시스템) 3개만 지원한다. 

matrix 에도 나와 있지만 OCFS 는 원래 구성을 지원하지 않는다.

또한 infoscale 의 최신 matrix 에도 아직 26ai 는 지원되지 않는다. SFRAC 지원되지 않음.

ASM+CFS 혼용해서 같이 사용할 수도 있겠지만...

하지만 하지 말라고 하면 더 하고 싶어지는 법!!!

- ASM 으로 구성한 후에 OCFS 로 바꾸어서 공유영역으로 사용해 보자.

[root@ocfs1 ~]# df -h
Filesystem                 Size  Used Avail Use% Mounted on
devtmpfs                   4.0M     0  4.0M   0% /dev
tmpfs                      4.9G     0  4.9G   0% /dev/shm
tmpfs                      2.0G  8.9M  2.0G   1% /run
/dev/mapper/ol_ocfs1-root   11G  5.5G  5.5G  51% /
/dev/sda1                  960M  427M  534M  45% /boot
/dev/mapper/ol_ocfs1-u01    40G  318M   40G   1% /u01
/dev/sdb                    20G  1.1G   19G   6% /oradata
/dev/sdc                    10G  551M  9.5G   6% /archive
/dev/sdd                   2.0G  279M  1.8G  14% /vote01
/dev/sdf                   2.0G  279M  1.8G  14% /vote02
/dev/sde                   2.0G  279M  1.8G  14% /vote03
tmpfs                      995M   36K  995M   1% /run/user/0

현재는 공유영역으로 사용하기 위해 이렇게 구성이 되어 있다.

- 우선 디스크를 하나 더 붙이고 ASM 을 이용해서 VOTING FILE 이 저장될  디스크그룹을 만들어 준다.

만들어진 모습

[root@ocfs1]# crsctl stat res -t
--------------------------------------------------------------------------------
Name           Target  State        Server                   State details       
--------------------------------------------------------------------------------
Local Resources
--------------------------------------------------------------------------------
ora.LISTENER.lsnr
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
ora.chad
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
ora.helper
               OFFLINE OFFLINE      ocfs1                    ILDE,STABLE
               OFFLINE OFFLINE      ocfs2                    IDLE,STABLE
ora.net1.network
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
ora.ons
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
--------------------------------------------------------------------------------
Cluster Resources
--------------------------------------------------------------------------------
ora.ASMNET1LSNR_ASM.lsnr(ora.asmgroup)
      1        ONLINE  ONLINE       ocfs1                    STABLE
      2        ONLINE  ONLINE       ocfs2                    STABLE
ora.LISTENER_SCAN1.lsnr
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.VOTE.dg(ora.asmgroup)
      1        ONLINE  ONLINE       ocfs1                    STABLE
      2        ONLINE  ONLINE       ocfs2                    STABLE
ora.asm(ora.asmgroup)
      1        ONLINE  ONLINE       ocfs1                    Started,STABLE
      2        ONLINE  ONLINE       ocfs2                    Started,STABLE
ora.asmnet1.asmnetwork(ora.asmgroup)
      1        ONLINE  ONLINE       ocfs1                    STABLE
      2        ONLINE  ONLINE       ocfs2                    STABLE
ora.cdp1.cdp
      1        OFFLINE OFFLINE                               STABLE
ora.cvu
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.cvuhelper
      1        OFFLINE OFFLINE                               STABLE
ora.ocfs1.vip
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.ocfs2.vip
      1        ONLINE  ONLINE       ocfs2                    STABLE
ora.rhpserver
      1        OFFLINE OFFLINE                               STABLE
ora.scan1.vip
      1        ONLINE  ONLINE       ocfs1                    STABLE
--------------------------------------------------------------------------------

[root@ocfs1 ~]# crsctl query css votedisk
##  STATE    File Universal Id                File Name Disk group
--  -----    -----------------                --------- ---------
 1. ONLINE   eada12b7b7504fdebfbb63aa9feb71d5 (/dev/sdg1) [VOTE]
Located 1 voting disk(s).

[root@ocfs1 ~]# ocrcheck
Status of Oracle Cluster Registry is as follows :
         Version                  :          4
         Total space (kbytes)     :     901284
         Used space (kbytes)      :      84848
         Available space (kbytes) :     816436
         ID                       : 1972206939
         Device/File Name         :      +VOTE
                                    Device/File integrity check succeeded
                                    Device/File not configured
                                    Device/File not configured
                                    Device/File not configured
                                    Device/File not configured
         Cluster registry integrity check succeeded
         Logical corruption check succeeded
 
- OCR과 VOTE 화일이 모두 +VOTE 디스크그룹에 저장되어 있다.

- 이제 공유영역으로 바꾸어 보자.
 
▣ VOTE 화일 이전

[root@ocfs1 ~]# crsctl replace votedisk /vote01/vote01
Now formatting voting disk: /vote01/vote01.
CRS-4256: Updating the profile
Successful addition of voting disk b73aa1ddef1c4f77bfcd2ecca87f789b.
Successful deletion of voting disk eada12b7b7504fdebfbb63aa9feb71d5.
CRS-4256: Updating the profile
CRS-4266: Voting file(s) successfully replaced

[root@ocfs1 ~]# crsctl add css votedisk /vote02/vote02
Now formatting voting disk: /vote02/vote02.
CRS-4603: Successful addition of voting disk /vote02/vote02.

[root@ocfs1 ~]# crsctl add css votedisk /vote03/vote03
Now formatting voting disk: /vote03/vote03.
CRS-4603: Successful addition of voting disk /vote03/vote03.

[root@ocfs1 ~]# crsctl query css votedisk
##  STATE    File Universal Id                File Name Disk group
--  -----    -----------------                --------- ---------
 1. ONLINE   b73aa1ddef1c4f77bfcd2ecca87f789b (/vote01/vote01) []
 2. ONLINE   6cffa04aaef44fc5bfe8689f37173c6c (/vote02/vote02) []
 3. ONLINE   a354031602514fdebf622bc1499ed207 (/vote03/vote03) []
Located 3 voting disk(s).

▣ OCR 화일 이전

[root@ocfs1 ~]# ocrconfig -add /vote01/ocr01
PROT-30: The Oracle Cluster Registry location to be added is not usable.
PROC-50: The Oracle Cluster Registry location to be added is inaccessible on nodes ocfs1, ocfs2.

화일이 없어서 에러가 난다. 빈 화일을 만들어 준다.

touch /vote01/ocr01 /vote01/ocr01 /vote01/ocr01
chown oracle:dba /vote01/ocr01 /vote02/ocr02 /vote03/ocr03
chmod 640 /vote01/ocr01 /vote02/ocr02 /vote03/ocr03 (voting file 인 경우는 chmod 644)

[root@ocfs1 vote01]# ocrconfig -add /vote01/ocr01
[root@ocfs1 vote01]# ocrconfig -add /vote02/ocr02
[root@ocfs1 vote01]# ocrconfig -add /vote03/ocr03

[root@ocfs1 vote03]# ocrcheck
Status of Oracle Cluster Registry is as follows :
         Version                  :          4
         Total space (kbytes)     :     901284
         Used space (kbytes)      :      84848
         Available space (kbytes) :     816436
         ID                       : 1972206939
         Device/File Name         :      +VOTE
                                    Device/File integrity check succeeded
         Device/File Name         : /vote01/ocr01
                                    Device/File integrity check succeeded
         Device/File Name         : /vote02/ocr02
                                    Device/File integrity check succeeded
         Device/File Name         : /vote03/ocr03
                                    Device/File integrity check succeeded

                                    Device/File not configured

         PROT-607: More than one OCR storage type was detected.

         Cluster registry integrity check succeeded

         Logical corruption check succeeded
 
OCR 화일도 정상 추가 되었다.

▣ 남아 있는 ASM 정보를 삭제하자. 일단 삭제하려고 하는데 기본 백업위치가 ASM 이라 지워지지 않는다.
  
[root@ocfs1 vote03]# ocrconfig -delete +VOTE
PROT-48: The Oracle Cluster Registry location +VOTE cannot be deleted because the OCR backup location is an ASM disk group.

[root@ocfs1 vote03]# ocrconfig -showbackup

ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/backup00.ocr.258.1242819061     2505374838
ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/day.ocr.259.1242819065     2505374838
ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/week.ocr.260.1242819065     2505374838
PROT-25: Manual backups for the Oracle Cluster Registry are not available

[root@ocfs1 vote03]# ocrconfig -showbackuploc
The Oracle Cluster Registry backup location is [+VOTE].

▣ 백업 Location 을 새로 구성한 곳으로 지정해 준다.

[root@ocfs1 vote03]# ocrconfig -backuploc /vote01/
[root@ocfs1 vote03]# ocrconfig -showbackuploc
The Oracle Cluster Registry backup location is [/vote01/].

▣  이제 삭제가 된다.

[root@ocfs1 vote03]# ocrconfig -delete +VOTE
[root@ocfs1 vote03]# ocrconfig -showbackup

ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/backup00.ocr.258.1242819061     2505374838
ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/day.ocr.259.1242819065     2505374838
ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/week.ocr.260.1242819065     2505374838

▣ 수동으로 백업을 한번 받아준다.

[root@ocfs1 vote03]# ocrconfig -manualbackup
ocfs1     2026/09/01 12:40:40     /vote01/backup_20260901_124040.ocr     2505374838     

[root@ocfs1 vote03]# ocrconfig -showbackup

ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/backup00.ocr.258.1242819061     2505374838
ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/day.ocr.259.1242819065     2505374838
ocfs1     2026/09/01 11:31:03     +VOTE:/ocfs-cluster/OCRBACKUP/week.ocr.260.1242819065     2505374838
ocfs1     2026/09/01 12:40:40     /vote01/backup_20260901_124040.ocr     2505374838  

▣ 만일 이전 정보가 남아 있어서 깔끔하게 정리하고 싶다면 ASM 삭제전에 백업화일 들도 정리해 주자 (이건 안해봐서 모르겠다)

ocrconfig -delbackup +VOTE:/ocfs-cluster/OCRBACKUP/backup00.ocr.258.1242819061
ocrconfig -delbackup +VOTE:/ocfs-cluster/OCRBACKUP/day.ocr.259.1242819065
ocrconfig -delbackup +VOTE:/ocfs-cluster/OCRBACKUP/week.ocr.260.1242819065

▣ OCR 화일 확인

[root@ocfs1 vote03]# ocrcheck
Status of Oracle Cluster Registry is as follows :
         Version                  :          4
         Total space (kbytes)     :     901284
         Used space (kbytes)      :      84880
         Available space (kbytes) :     816404
         ID                       : 1972206939
         Device/File Name         : /vote01/ocr01
                                    Device/File integrity check succeeded
         Device/File Name         : /vote02/ocr02
                                    Device/File integrity check succeeded
         Device/File Name         : /vote03/ocr03
                                    Device/File integrity check succeeded
                                    Device/File not configured
                                    Device/File not configured
         Cluster registry integrity check succeeded
         Logical corruption check succeeded


▣ 이제 남아 있던  ASM Resource 의 흔적을 깨끗이 삭제해 준다.

srvctl update listener -listener ASMNET1LSNR_ASM -asm -remove -force
PRCR-1065 : Failed to stop resource ora.ASMNET1LSNR_ASM.lsnr
CRS-2670: Unable to start/relocate 'ora.asm_listener.type' because 'ora.asm' has a stop-time 'hard' dependency on it
CRS-5027: The ASM instance on node "ocfs1" cannot be stopped because it is currently being used by Cluster Ready Services.

-> 아직 ASM 에 정보가 남아서 인지 삭제되지 않는다. 재기동 해보자.  

crsctl stop crs -f
crsctl start crs

다시 진행

srvctl update listener -listener ASMNET1LSNR_ASM -asm -remove -force
srvctl stop asmnetwork
srvctl remove asmnetwork -all

srvctl stop diskgroup -g VOTE
srvctl remove diskgroup -diskgroup VOTE -force
srvctl stop asm -node ocfs1
srvctl stop asm -node ocfs2
srvctl remove asm -f

crsctl stop res ora.asm -init     
crsctl delete res ora.asm -init  <--- CRS daemon 도 기동되지 않게 삭제한다.

▣ 상태 확인
[root@ocfs1 ~]# crsctl stat res -t
--------------------------------------------------------------------------------
Name           Target  State        Server                   State details       
--------------------------------------------------------------------------------
Local Resources
--------------------------------------------------------------------------------
ora.LISTENER.lsnr
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
ora.chad
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
ora.helper
               OFFLINE OFFLINE      ocfs1                    IDLE,STABLE
               OFFLINE OFFLINE      ocfs2                    IDLE,STABLE
ora.net1.network
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
ora.ons
               ONLINE  ONLINE       ocfs1                    STABLE
               ONLINE  ONLINE       ocfs2                    STABLE
--------------------------------------------------------------------------------
Cluster Resources
--------------------------------------------------------------------------------
ora.LISTENER_SCAN1.lsnr
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.cdp1.cdp
      1        OFFLINE OFFLINE                               STABLE
ora.cvu
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.cvuhelper
      1        OFFLINE OFFLINE                               STABLE
ora.ocfs1.vip
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.ocfs2.vip
      1        ONLINE  ONLINE       ocfs2                    STABLE
ora.rhpserver
      1        OFFLINE OFFLINE                               STABLE
ora.scan1.vip
      1        ONLINE  ONLINE       ocfs1                    STABLE
--------------------------------------------------------------------------------

[root@ocfs1 ~]# crsctl stat res -t -init
--------------------------------------------------------------------------------
Name           Target  State        Server                   State details       
--------------------------------------------------------------------------------
Cluster Resources
--------------------------------------------------------------------------------
ora.cluster_interconnect.haip
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.crf
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.crsd
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.cssd
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.cssdmonitor
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.diskmon
      1        OFFLINE OFFLINE                               STABLE
ora.drivers.acfs
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.evmd
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.gipcd
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.gpnpd
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.mdnsd
      1        ONLINE  ONLINE       ocfs1                    STABLE
ora.storage
      1        ONLINE  ONLINE       ocfs1                    STABLE
--------------------------------------------------------------------------------

[root@ocfs1 ~]# df -h
Filesystem                 Size  Used Avail Use% Mounted on
devtmpfs                   4.0M     0  4.0M   0% /dev
tmpfs                      4.9G     0  4.9G   0% /dev/shm
tmpfs                      2.0G   13M  2.0G   1% /run
/dev/mapper/ol_ocfs1-root   11G  6.1G  4.9G  56% /
/dev/sda1                  960M  427M  534M  45% /boot
/dev/mapper/ol_ocfs1-u01    40G  5.0G   35G  13% /u01
/dev/sde                   2.0G  495M  1.6G  25% /vote03
/dev/sdd                   2.0G  495M  1.6G  25% /vote01
/dev/sdf                   2.0G  495M  1.6G  25% /vote02
/dev/sdc                    10G  551M  9.5G   6% /archive
/dev/sdb                    20G  1.1G   19G   6% /oradata
tmpfs                      995M   36K  994M   1% /run/user/0
tmpfs                      995M   36K  994M   1% /run/user/1001
oracle_clusterware         128M  4.7M  124M   4% /u01/app/oracle/crsdata/ocfs1/shm

공유영역에 DB 생성하는것 까지는 귀찮아서 안해봤지만 생성은 잘 될것 같다.

작업 끝~~

26ai 23.26.3 버전부터 LOG_REDO_PRIORITIZATION (자동 Redo 우선순위 지정) 라는 새로운 파라미터가 도입되었다.

https://blogs.oracle.com/maa/auto-redo-prioritization-helping-manage-redo-generation-spikes

SQL> show parameter log_redo_prioritization

NAME                                 TYPE        VALUE
------------------------------------ ----------- ------------------------------
log_redo_prioritization              boolean     FALSE

기본값은 FALSE

SQL> ALTER SYSTEM SET LOG_REDO_PRIORITIZATION = TRUE SCOPE=BOTH;
재기동 필요없이 동적으로 적용가능함.

기본적으로는 Primary 에 설정, 하지만 switchover 되는 상황을 고려한다면 양쪽 다 셋팅.

: 대량의 DML이나 배치 작업으로 인해 Redo Log Buffer 경합이 발생할 때, OLTP 트랜잭션의 지연을 방지하기 위해 배치 세션의 Redo 생성을 일시적으로 조절하는 역할을 한다.

- 동작원리 및 매커니즘

1. 버퍼 사용률 모니터링 : 파라미터를 TRUE로 설정하면 오라클 인스턴스가 Redo Log Buffer의 여유 공간을 실시간으로 감시.

2. 자동 워크로드 분류 :  관리자가 Resource Manager 컨슈머 그룹을 사전에 수동 매핑하지 않아도, 데이터베이스가 세션의 Redo 발생량 및 호출 지속 시간(Call Duration) 등을 분석하여 OLTP 성격과 배치(Batch) 성격 세션을 자체 판별.

3. 선별적 쓰로틀링(Throttling) : 여유 공간이 내부 임계치 미만으로 떨어지면, 배치 세션에 대기를 유발하여 Redo 생성을 지연시킴. 이때 지연된 세션은 redo prioritization for log write lag 대기 이벤트를 기록.

4. OLTP 세션 보호 : Redo Log Buffer 여유 공간을 Latency에 민감한 OLTP 세션이 우선 선점하도록 하여 log buffer space 대기 발생을 최소화합니다.

우선순위는 OLTP가 동작하는 세션에는 영향을 주지 않으며, 배치 동작을 하는 세션에만 영향을 준다.

여기서, LOG_REDO_PRIORITIZATION의 내부 임계값은 사용자가 수동으로 지정하는 고정된 단일 수치(예: 80% 고정 등)가 아니라, 커널 내부 알고리즘에 의해 실시간 동적(Dynamic)으로 계산되는 적응형 임계치를 사용함.

- 오라클 백그라운드 프로세스(LGWR 및 워커 프로세스)가 판단에 활용하는 핵심 기준 지표는 다음과 같음.

1. Redo Log Buffer 가용 임계치 (Buffer Headroom)

동적 가용률 감시: Redo Log Buffer 전체 크기 중 미기록된 Redo가 차지하는 비중을 실시간 추적.

LGWR I/O 처리율 대비 생성 속도: 버퍼의 단순 점유율뿐만 아니라, LGWR의 초당 디스크 Write 처리량(Redo Write Rate) 대비 현재 전체 세션의 Redo 생성 속도(Redo Generation Rate)의 비율이 1을 초과하여 급격한 버퍼 소진(log buffer space 발생 위험)이 예측되는 시점을 임계 상태로 판별.

2. 세션별 Redo 생성률 임계치 (Batch vs. OLTP 분류 기준)

단위 호출당 Redo 생성량(Redo per Call / Transaction): 짧은 트랜잭션(일반적인 OLTP)과 달리, 단일 DB Call 또는 단위 시간(초) 동안 대량의 Redo Vector를 지속적으로 쏟아내는 세션을 시스템 통계(AWR 기준 상위 p95/p99 생성 패턴)와 비교하여 배치 세션으로 동적 마킹함

지속 시간(Burst Duration): 일시적인 짧은 스파이크가 아닌, 수 초 이상 지속적으로 높은 Redo 발생량을 유지하는 세션이 쓰로틀링 대상이 됨.

3. Data Guard 연계 시의 전송 지연 임계치 (Transport Lag Threshold)

LOG_ARCHIVE_DEST_n의 REDO_PRIORITIZATION_WHEN_TRANSPORT_LAGS_BY_SEC 옵션과 연계된 경우, 이 파라미터에 명시된 초(Seconds) 단위의 전송 지연 시간이 절대 임계값으로 작용합니다. (예: 30초 이상 Lag 발생 시 배치 쓰로틀링 강도 급증)

ADG 환경이라면 이렇게 활용할 수 있다.

ALTER SYSTEM SET LOG_ARCHIVE_DEST_3 =  'SERVICE=boston ASYNC NOAFFIRM   VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE)   DB_UNIQUE_NAME=boston REDO_PRIORITIZATION_WHEN_TRANSPORT_LAGS_BY_SEC=20';

DG broker 를 사용한다면..

EDIT DATABASE newyork SET PROPERTY
   AdditionalLadAttributes='REDO_PRIORITIZATION_WHEN_TRANSPORT_LAGS_BY_SEC=20';

->  우선순위 지정으로 Async physical standby destination 에  대한 목표 전송 지연을 정의합니다. 전송 지연이 목표를 초과하면 재전송이 자원 관리자에게 알림을 보내고, 자원 관리자가 재 우선순위를 적용한다.