개발 공부
(Oracle, dataguard) sync remote write event 본문
728x90
"sync remote write" means:
A foreground session is waiting for redo to be written to a remote destination (typically a Data Guard standby database) in synchronous mode.
📦 When Does It Happen?
- When you're using Oracle Data Guard in SYNC mode (i.e., SYNC AFFIRM or SYNC NOAFFIRM).
- The primary database must wait until the redo is written to the standby's standby redo log (SRL).
- This happens during commits, DML, or other redo-generating operations.
⛓ Why the Wait?
Oracle ensures zero data loss in SYNC mode — so:
- Primary sends redo to the standby.
- Primary waits for acknowledgment:
- AFFIRM: Waits until redo is flushed to disk on the standby.
- NOAFFIRM: Waits until redo is received in memory on the standby.
Only after that, the primary proceeds — which causes the "sync remote write" wait event.
✅ "sync remote write" Is Associated with LGWR
🔍 Why?
When you're using Data Guard in SYNC mode, Oracle uses the Log Writer (LGWR) process on the primary database to transmit redo to the standby.
- This is different from ASYNC mode, which uses LNS (Log Network Server).
- In SYNC mode, LGWR is responsible for writing to both:
- The local online redo log
- The remote standby redo log (over the network)
That’s why:
- The wait event is "sync remote write" — the foreground session (e.g., a user COMMIT) waits until LGWR confirms that the redo was flushed to the standby (or at least acknowledged).
- You see the wait event tied to a foreground session, but the action being waited on is performed by LGWR.
728x90
'DB, SQL' 카테고리의 다른 글
(Oracle, datagaurd) sync mode, affirm 모드 확인 방법 (0) | 2025.05.19 |
---|---|
(Oracle, dataguard) 쓰기 지연 현상 파악 (0) | 2025.05.19 |
(Oracle) 데이터 삭제 시 인덱스 공간 회수 (1) | 2024.11.20 |
인덱스 리빌딩과 테이블 스페이스 (1) | 2024.11.20 |
테이블스페이스 증가 시키기 (0) | 2024.11.20 |
Comments