Start/notes.ini Parameters/TRANSLOG_Performance

TRANSLOG_Performance

Parameter: TRANSLOG_Performance
Short description: Determines the trade-off between runtime performance and restart/recovery time for transaction logging: 1 = Favor runtime, 2 = Standard (default), 3 = Favor restart recovery time. Corresponds to the "Runtime/Restart Performance" field in the server document.

Profile

Parameter
TRANSLOG_Performance
Component
Server
Category
Transaction log
Available since
Domino 5.x; currently documented in HCL Domino 9.0.x–14.5.1
Value format
Integer 1, 2, or 3
Default
2 (Standard)
GUI equivalent
Server document → "Transactional Logging" → "Runtime/Restart Performance"

Description

Domino writes transactions sequentially to the translog first but holds the changed NSF blocks in the Unified Buffer Manager (UBM) in memory. These blocks are not hardened to the NSF file until the next checkpoint. The more is held in the UBM, the faster the runtime – but the more must be replayed from the log after a crash.
Per HCL KB0032176 "Transactional Logging and how it operates", there are three levels:

TRANSLOG_Performance=1 – Favor runtime

  • Redo.Limit = min(500 MB, half the defined log size). For a 4 GB log size, e.g. 2 GB Redo.Limit (capped at 500 MB).
  • Few checkpoints, much UBM content, maximum runtime performance.
  • Restart recovery takes a long time, since correspondingly many transactions must be replayed from the log into the NSFs.
  • Suitable for large servers with plenty of RAM and stable hardware/cluster environments.

TRANSLOG_Performance=2 – Standard (default)

  • Redo.Limit = 49 MB.
  • Balanced relationship between runtime performance and restart recovery.
  • Recommended default for almost all production servers.

TRANSLOG_Performance=3 – Favor restart recovery time

  • Many checkpoints, little UBM content, fast restart.
  • Runtime slightly slower because hardening to NSF happens more frequently.
  • Suitable for high-availability clusters where a fast restart is more important than peak-load performance.

Examples

Standard recommendation for typical production servers:
TRANSLOG_Status=1
TRANSLOG_Style=1
TRANSLOG_Path=D:\Domino\txn
TRANSLOG_Performance=2
Large RAM server, where peak load matters more than restart time:
TRANSLOG_Status=1
TRANSLOG_Style=1
TRANSLOG_Path=D:\Domino\txn
TRANSLOG_Performance=1
HA cluster node focused on fast recovery:
TRANSLOG_Status=1
TRANSLOG_Style=1
TRANSLOG_Path=D:\Domino\txn
TRANSLOG_Performance=3

Notes

  • Effective only after restart – TRANSLOG_Performance takes effect at the next server start.
  • Watch RAM – "Favor runtime" (1) can be counterproductive in tight memory situations if the UBM grows too large and consumes the OS page file.
  • Translog vs. NSF disk performance – If frequent hardening (3) is chosen, NSF I/O must also be smooth, otherwise hardening eats up the gained restart time.
  • Companion parameters – TRANSLOG_Status, TRANSLOG_Path, TRANSLOG_Style, TRANSLOG_MaxSize, TRANSLOG_AutoFixup, TRANSLOG_UseAll.

Sources (HCL Product Documentation)