Changing tunable values for Asynchronous I/O
All of the AIO tunables have current, default, minimum and maximum values that can be viewed with the ioo command.
Only the current value of a tunable can be changed with the ioo command. The other three
values are fixed and are presented to inform you of the bounds of
the tunable. You can change the current value of the tunable anytime
and make it persistent for the restart of the operating system. All
of the tunables follow the normal rules and options controlled by
the ioo command, which is in the performance tool
file set.
Nonrestricted tunables are summarized in the following table:
Item | Description |
---|---|
minservers |
Indicates the minimum number of kernel processes
per processor dedicated to AIO processing. Because each kernel process
uses memory, the minservers tunable value, when multiplied
by the number of processors, must not be large when the amount of
AIO expected is small. The default value for the minservers tunable
is 3. |
maxservers |
Indicates the maximum number of kernel processes
per processor that are dedicated to AIO processing. This tunable value,
when multiplied by the number of processors, indicates the limit on
slow path I/O requests in progress at one time and represents the
limit for possible I/O concurrency. The default value for the maxservers tunable
is 30. |
maxreqs |
Indicates the maximum number of AIO requests
that can be outstanding at one time. The requests include those that
are in progress, as well as those that are waiting to be started.
The maximum number of AIO requests cannot be less than the value of AIO_MAX ,
as defined in the /usr/include/sys/limits.h file,
but it can be greater. It is appropriate for a system with a high
volume of AIO to have a maximum number of AIO requests larger than AIO_MAX .
The default value for the maxreqs tunable is 16384. |
server_inactivity |
Indicates a time-out value, in seconds, that
causes the server to exit if it is idle (sleeping) without servicing
an AIO request. If exiting would cause the total number of servers
to fall below minservers * number of CPUs, the server
goes back to sleep waiting for an AIO to service. This mechanism helps
overall system performance by reducing the number of sleeping processes
that are not being used to service AIO requests. The default value
for the server_inactivity tunable is 300. |