DISK_ASYNCH_IO - Enable Oracle Asynchronous I/O

DISK_ASYNCH_IO

Overview

DISK_ASYNCH_IO controls whether Oracle uses asynchronous I/O for disk operations against datafiles, control files, and log files stored on file systems. When enabled (the default), Oracle submits I/O requests to the operating system and continues processing without waiting for each request to complete — the OS notifies Oracle when each I/O finishes. This allows database writer processes (DBWn) and other background processes to overlap I/O with other work, improving throughput especially on systems with high write concurrency.

When DISK_ASYNCH_IO is set to FALSE, Oracle falls back to synchronous I/O — each I/O request blocks the calling process until it completes. This can serialize I/O and significantly reduce throughput on write-intensive workloads. This parameter applies only to file system files; ASM disk group I/O uses a separate asynchronous mechanism and is not affected by this setting.

Parameter Type: Static (requires instance restart to change) Default Value: TRUE Valid Values: TRUE, FALSE Available Since: Oracle 8i Modifiable: No — SCOPE=SPFILE only; takes effect on next startup PDB Modifiable: No (CDB-level setting)

Configuration

Viewing Current Value

-- Check current DISK_ASYNCH_IO setting
SELECT name, value, isdefault, ismodified, description
FROM v$parameter
WHERE name = 'disk_asynch_io';

-- Check SPFILE value (persistent setting for next startup)
SELECT name, value, isspecified
FROM v$spparameter
WHERE name = 'disk_asynch_io';

-- Check related async I/O parameters together
SELECT name, value
FROM v$parameter
WHERE name IN (
  'disk_asynch_io',
  'tape_asynch_io',
  'filesystemio_options',
  'db_writer_processes'
)
ORDER BY name;

-- Check whether async I/O is actually being used (OS-level view)
SELECT name, value
FROM v$parameter
WHERE name = 'disk_asynch_io';

Setting the Parameter

-- Enable asynchronous I/O (default; recommended)
ALTER SYSTEM SET disk_asynch_io = TRUE SCOPE=SPFILE;

-- Disable asynchronous I/O (for diagnosis or unsupported OS configurations)
ALTER SYSTEM SET disk_asynch_io = FALSE SCOPE=SPFILE;

-- After changing, restart the instance to apply
-- SHUTDOWN IMMEDIATE;
-- STARTUP;

-- Verify the new setting after restart
SELECT name, value, isdefault
FROM v$parameter
WHERE name = 'disk_asynch_io';

Tuning Guidance

Recommended Values

Environment	Recommended Setting
Linux (ext4, XFS, NFS with async)	TRUE (default)
AIX with JFS2	TRUE
Solaris with UFS	TRUE
Windows NTFS	TRUE
NFS without async mount option	FALSE or use FILESYSTEMIO_OPTIONS
Legacy OS without kernel async I/O	FALSE
ASM disk groups (all platforms)	Not applicable — ASM uses its own async mechanism

In the vast majority of modern production environments, DISK_ASYNCH_IO = TRUE is correct and should not be changed. Disabling it is only appropriate when diagnosing I/O-related problems or when the OS configuration does not support kernel-level async I/O.

How to Size

This is a boolean parameter with no numeric tuning. The key decision is whether your OS and file system combination supports asynchronous I/O properly. Verify OS-level async I/O support before relying on this feature.

-- Monitor I/O wait events to assess async I/O effectiveness
SELECT event, total_waits, time_waited,
       ROUND(time_waited / NULLIF(total_waits, 0), 2) AS avg_wait_ms
FROM v$system_event
WHERE event IN (
  'db file async I/O submit',
  'db file sequential read',
  'db file scattered read',
  'db file parallel read',
  'log file parallel write',
  'log file sync'
)
ORDER BY time_waited DESC;

-- Check DBWn efficiency (high async I/O benefit visible here)
SELECT name, value
FROM v$sysstat
WHERE name IN (
  'physical write total IO requests',
  'physical write total bytes',
  'DBWR checkpoint buffers written',
  'DBWR buffers scanned'
)
ORDER BY name;

-- Review background process wait patterns
SELECT p.program, e.event, e.total_waits, e.time_waited
FROM v$process p
JOIN v$session s ON p.addr = s.paddr
JOIN v$session_event e ON s.sid = e.sid
WHERE p.background = 1
  AND e.event LIKE '%I/O%'
ORDER BY e.time_waited DESC;

Monitoring

-- Check for I/O-related wait events that indicate async I/O issues
SELECT event, total_waits, total_timeouts,
       time_waited, average_wait
FROM v$system_event
WHERE event LIKE '%async%'
   OR event LIKE '%db file%'
ORDER BY time_waited DESC
FETCH FIRST 10 ROWS ONLY;

-- Examine I/O throughput by file
SELECT f.name, s.phyrds, s.phywrts,
       s.readtim, s.writetim,
       ROUND(s.writetim / NULLIF(s.phywrts, 0), 2) AS avg_write_ms
FROM v$datafile f
JOIN v$filestat s ON f.file# = s.file#
WHERE s.phywrts > 0
ORDER BY s.writetim DESC;

Common Issues

Issue 1: Async I/O Enabled but OS Does Not Support It

On some NFS mounts or legacy file systems, DISK_ASYNCH_IO = TRUE may be set but the OS silently falls back to synchronous I/O. Oracle will not report an error, but I/O performance will be degraded.

Resolution: Verify OS async I/O is enabled and working. On Linux, check that libaio is installed and that the mount options for NFS volumes include async or that FILESYSTEMIO_OPTIONS=ASYNCH is configured where appropriate.

-- After a suspected fallback, compare wait events before and after
-- disabling DISK_ASYNCH_IO to see if behavior changes
SELECT event, time_waited
FROM v$system_event
WHERE event IN ('db file async I/O submit', 'db file sequential read')
ORDER BY event;

Issue 2: I/O Errors After Disabling Async I/O

Setting DISK_ASYNCH_IO = FALSE for diagnosis and then forgetting to re-enable it leaves the database running in synchronous I/O mode permanently. This commonly surfaces as elevated db file sequential read or log file sync wait times.

Resolution: Re-enable the parameter and restart the instance. Monitor wait events after the restart to confirm improvement.

-- Confirm async I/O is re-enabled after restart
SELECT name, value
FROM v$parameter
WHERE name = 'disk_asynch_io';

-- Confirm I/O wait events return to baseline
SELECT event, total_waits, time_waited
FROM v$system_event
WHERE event IN ('db file sequential read', 'log file sync')
ORDER BY time_waited DESC;

Issue 3: Confusion with FILESYSTEMIO_OPTIONS

DISK_ASYNCH_IO is sometimes confused with FILESYSTEMIO_OPTIONS. The former is a boolean that enables or disables Oracle’s use of OS async I/O at a high level. The latter controls more granular options including direct I/O and is platform-specific. Both may need to be configured together for optimal performance.

Resolution: Review both parameters together and consult platform-specific Oracle documentation.

-- Review all I/O-related parameters at once
SELECT name, value, isdefault
FROM v$parameter
WHERE name IN ('disk_asynch_io', 'filesystemio_options', 'tape_asynch_io')
ORDER BY name;

Related Parameters

• FILESYSTEMIO_OPTIONS — Controls direct I/O and async I/O at the file system level; used alongside this parameter
• DB_WRITER_PROCESSES — Controls the number of database writer processes; async I/O amplifies DBWn throughput
• DB_FILE_MULTIBLOCK_READ_COUNT — Controls read I/O size for full scans; pairs with async I/O for sequential read throughput
• LOG_BUFFER — Redo log buffer size; log writer also benefits from async I/O when enabled

Related Errors

• ORA-27125: Unable to Create Shared Memory Segment — Can surface in conjunction with I/O misconfiguration on some platforms
• ORA-01578: ORACLE Data Block Corrupted — In rare cases, improper async I/O on unsupported NFS configurations has been linked to block corruption; always verify OS-level async I/O support

Version Notes

Version	Notes
Oracle 8i	Parameter introduced
Oracle 10g	Default TRUE; async I/O used by DBWn, LGWR, and CKPT
Oracle 11g	No functional changes; behavior consistent across supported platforms
Oracle 12c+	Fully supported; ASM continues to use its own separate async I/O path
Oracle 19c+	No changes; TRUE remains the correct default for all modern OS configurations
Oracle 21c / 23ai	Behavior unchanged; verify NFS mount options when using dNFS (Direct NFS) alongside this parameter